Professional Documents
Culture Documents
max
Maximum inelastic displacement
Material Ducility
r
(t) Time varying velocity of a SDOF relative to the base of the structure
r
(t) Time varying acceleration of a SDOF relative to the base of the structure
g
(t) Time varying acceleration of the ground
Natural frequency of vibration of a structure
M Mass of a structure
F(t) Applied time varying load
f
I
(t) Inertial force
f
D
(t) Damping force
xxi
f
S
(t) Elastic force
F
e
Elastic strength demand
F
y
Yield strength demand
F
s
Design strength demand
C
d
Displacement amplification factor used in design
R
e
Displacement from a linear analysis
f
cu
Characteristic cube strength of concrete
f
c
Characteristic cylinder strength of concrete
f
t
Tensile strength of concrete
K Confinement factor of concrete
f
cc
Confined cylinder strength of concrete
F
y
Yield strength of reinforcing steel
F
u
Ultimate strength of reinforcing steel
E Youngs modulus
A Area
F
b
Design base shear force
W Weight of structure used in determining base shear
C
w
Base shear ratio
S
a
(T) Spectral acceleration on a design response spectrum
1
Importance factor of a structure according to table 8-9, ECP-201(2008)
E
Live load factor according to ECP-201
I
g
Gross moment of inertia of the cross-section
xxii
I
eff
Effective moment of inertia of the cross-section
Mass proportional damping coefficient (or alternatively, post-yield stiffness ratio)
Stiffness proportional damping coefficient
M
s
Richter scale earthquake magnitude
D
s
Earthquake epi-central distance
T
p
Earthquake predominant period
T
n
Period of the structure for the n
th
mode
T
B
Soil parameter
T
C
Soil parameter
T
D
Soil parameter
S Soil parameter
IDei Inter-storey drift ratio from elastic equivalent static analysis at floor i
dei Displacement from elastic equivalent static analysis at floor i
dsi Code estimated maximum displacement at floor i
ei Maximum elastic displacement from THA at floor i
maxii Maximum inelastic displacement from THA at floor i
IDi Maximum inter-storey drift ratio from inelastic analysis at floor i
Chapter 1
CHAPTER 1
INTRODUCTION
1.1 GENERAL INTRODUCTION
Severe damage of buildings is the most distressing consequence of most natural hazards,
especially earthquakes. In the past 20 years, Egypt was struck by five major earthquakes
namely the 1992 Cairo earthquake, the 1993 and 1995 three Gulf of Aqaba earthquakes and
the 1998 Alexandria earthquake (Abou Elenean et al., 2008). Although the 1995 Gulf of
Aqaba was the strongest of them, it is the earthquake of 1992 that has left the deepest
imprints on everyone, not only because it resulted in hundreds killed and injured, but also
because it incurred an economic loss of approximately 4 billion Egyptian pounds in damages,
making it one of the costliest natural disasters in Egypt.
In the past, limiting damage was not a prime objective of seismic design codes which
concentrated only on providing an adequate level of life safety with the justification that
earthquakes are generally unpredictable and tend to be rare compared to other lateral loads
such as wind. Recent observations of actual behavior of buildings during some strong
earthquakes have demonstrated that this philosophy has some serious shortcomings where in
many structures, which have satisfied the no-collapse requirement, the damage inflicted has
resulted in vast economic loss, high cost of repairs, as well as indirect costs due to business
interruptions, need for relocation of inhabitants, etc. The lesson learnt was that these costs are
prohibitive, and thus, even if the events are rare, seismic codes should target mitigating
damages as well as protecting life safety. Therefore most recent seismic codes, including
2008 draft of the Egyptian Code for Design Loads on Structures (ECP-201, 2008), are
developed with two performance levels, one with the intent of limiting damage during
frequent moderate earthquakes namely the serviceability limit state and the other ensuring
collapse prevention during a major earthquake namely the ultimate limit state.
1
Chapter 1
Since damage sustained by a structure depends on the amount of deformations it
experiences, displacement parameters offer better evaluation of damage effects than force
parameters when assessing structures to a serviceability limit state (Priestley, 1997). A
building is considered in a damage state if it has reached the yielding point on a force-
displacement diagram. For reinforced concrete structures, yielding and cracking takes place
at quite low displacements. After yielding, there is considerable increase in displacement as
compared to the increase in forces, thus the structure suffers significant progressive damage
with a small increase in force. Therefore limiting displacement is a requirement for
controlling the damage.
Moreover, earthquake engineering research has proven that deformability is a governing
factor in satisfying seismic codes life-safety requirement as well as damage limitation
((FEMA 451B-7, 2003). During the 1994 Northridge Earthquake a strong earthquake that
struck a neighborhood in the city of Los Angeles in California and resulted in an estimated
damage of $20 billion leading to major rethinking of seismic design practice at that time
many well-designed and detailed buildings collapsed because their structural elements could
not all deform with the building during the event, and in some other buildings certain
nonstructural elements that did not have the capacity to deform got damaged and fell off
killing passers-by and obstructing occupants escape from the buildings, thus defying the life-
safety objective they were designed for (Taranath, 2005). Therefore in seismic design a
philosophy evolved that deformability should be controlled by assessing elements
deformation capacity relative to the deformation demands which made estimating the
maximum displacement of structures a life-safety criterion that cannot be skipped by
designers.
1.1.1 Definition of drift
An earthquake principally causes movement of the ground and hence a base
displacement rather than a surface force like wind. As the ground moves in one direction, the
building base attached to it slides in the direction of movement, while the structure above the
ground lags behind because the inertia of the building mass resists movement. The amount of
horizontal displacement of any point in the building relative to the ground is called drift. Drift
causes stress in structural and nonstructural elements because it causes distortion. Because
2
Chapter 1
earthquake movement is a reversed cyclic motion, vibration is induced in the building and a
wave of distortion travels along the height of the building (Taranath, 2005). Maximum drift
usually takes place at the top of a building, but each story level is subjected to a certain
amount of drift known as inter-story drift, which is calculated as the difference between the
lateral displacements of one floor and the one below as shown in Figure 1.1.
Figure 1.1 Illustration of the definition of drift and storey drift
(ATC/SEAOC Joint Venture)
1.1.2 Effect of drift on structures
Without proper control during the design process, drift can have the following adverse
effects on a structure:
Effect on structural integrity
Many observations and dynamic-response studies have shown that a buildings collapse
potential is greatly affected by the amount of drift (Naeim, 1989). In the absence of
deformation compatibility, lateral displacement can affect both elements of the lateral force
resisting system and gravity load bearing elements, by causing too much distortion, leading to
3
Chapter 1
premature failure and loss of strength (Freeman et al., 2004). Also excessive drift can affect
the vertical stability of a building, especially flexible massive buildings, potentially leading to
collapse due to P- effects.
Effect on architectural elements
If nonstructural elements, for example partitions, glass and cladding systems, are not
well isolated from the movements of the structural elements, they may get damaged as a
result of drift and fall off jeopardizing safety of occupants and requiring high repair costs.
Also staircase enclosures tend to inhibit inter-story drift resulting in damage. Moreover, the
nonstructural elements may interfere with their attached structural elements and lead to some
undesirable results for example creating short columns, torsion, or stiffness irregularities
(Freeman et al., 2004).
Effect on adjacent Structures.
If neighboring buildings or structurally separate sections of the same building do not
have adequate separation, they may pound against each other during an earthquake. Pounding
may lead to irregular response of buildings especially if they are not of the same height,
resulting in local damage to columns as the floor of one building hits columns of another,
collapse of damaged floors, and possibly collapse of the whole building (Rosenblueth et al.,
1986).
1.1.3 Current seismic code approach for drift control
Since earthquakes impose deformations on structures and those deformations generate
stresses and bending in members, the logical design procedure would be to start with a
predefined target displacement and then progress with design to get the required strength and
stiffness under a certain design earthquake level. However, this displacement-based design
(DBD) method is currently in its infancy and not yet applied in seismic codes, while it is only
limited to research or rehabilitation of existing buildings, because without knowing structure
geometry, member dimensions and reinforcement, displacement demands are difficult to
4
Chapter 1
estimate to be the starting point of design. Also seismic design was originally developed as an
extension to gravity and wind load design, therefore it followed the same procedure which is
force-based (FBD), with controlling displacement coming just as a final check leading either
to changes in the structural system in an iterative process, or changes in detailing the structure
in order to change its ductility and capacity to deform (Borzi et al., 2000). It should be noted,
however, that the forces used in this conventional design practice are completely fictitious.
For estimating these design seismic forces, it has been proven that the assumption of
linear elastic behavior of structures during strong earthquakes would result in very high
structural costs in relation to the probability of occurrence of the maximum design
earthquake. Also since there is a lot of uncertainty about the magnitude of ground motions, it
can never be assured that, even if the structure is designed to behave elastically, the actual
forces will not exceed the adopted design forces by a large margin and thus entering the
inelastic range. Therefore, for common residential or commercial buildings, it is neither
practical nor economically feasible to design a structure to respond in the elastic range to a
ground motion representative of the maximum possible earthquake.
1
Buildings are designed
to behave elastically only to moderate earthquakes, and for strong earthquakes, the design
relies on the structures capacity to dissipate a substantial portion of the energy imparted to it
by inelastic action. This means that yielding is permitted thus requiring controlling damage in
the structure by prescribing special ductile detailing requirements in predetermined locations,
with the aim of creating a structure capable of sustaining post-yield displacements without
collapse. So the whole idea behind this concept is the understanding that a building designed
to behave nonlinearly should have a predictable ductile response which is much better than
the response of a building designed to behave elastically if it gets overloaded.
However, nonlinear analysis methods are too complex and onerous to be applied
systematically in the seismic design process, therefore, the new Egyptian code draft (ECP-
201, 2008), like many other building codes, uses the capacity of structural systems to resist
seismic actions in the inelastic range to allow performing an elastic analysis based on smaller
seismic forces than those corresponding to a linear elastic response. Thus the elastic force
demands are reduced by a force reduction factor (FRF) that represents the ductility behavior
1
With the exception of special high-performance buildings and buildings with special protective systems such
as nuclear facilities yielding cannot be tolerated and as such, the design needs to be elastic.
5
Chapter 1
of the structural system. ECP-201 (2008) uses the R-factor for this purpose namely the
Response Modification Factor (RMF); however this factor does not represent a modification
of a response per se, so the term FRF will be used in this study.
Since reduced seismic forces are used in the design, the computed displacement from
this elastic analysis is amplified by a factor herein called the displacement amplification
factor (DAF) in order to estimate the actual inelastic displacement
max
that develops in
strong earthquakes. ECP-201 specifies this DAF as equal to 0.7 multiplied by the FRF
assumed in the elastic analysis (2008). In other words, the code allows designing for drift
which is 70% of that that would happen in a theoretically elastic structure (SEAOC, 2009).
The resulting inelastic displacement is used in several checks throughout the code, namely: 1.
checking stability and safety of the structure by checking P effects; 2. defining
requirements for minimum buildings separation to avoid pounding; and 3. defining
limitation on inter-story drift to limit damage in case of a moderate frequent earthquake. Also
for achieving proper seismic resistance, designers should use
max
in checking deformation
capacity of critical structural members and in detailing connections for nonstructural
components (Uang, 1991).
1.2 PROBLEM STATEMENT
The use of the reduced-force-amplified-displacement method has served the seismic
design practice well with the benefits of simplicity and resulting reduced cost due to lower
design force, however the inherent weakness of the magnitudes assigned to the force
reduction (FRF) and displacement amplification (DAF) factors is widely accepted based on
many literature studies (ElNashai et al, 2002). The magnitudes of FRF and consequently the
DAF are specified by codes depending on the lateral-load resisting system and the material of
construction but these specified values are inconsistent and based on committee consensus,
therefore it is difficult to justify their value except by experiment or analysis. Specification of
a reliable DAF in seismic codes is essential since estimates of maximum inelastic
displacement are used in several checks during the seismic design process.
6
Chapter 1
In the past, there were no provisions in the Egyptian seismic code for drift limitations or
specifying minimum building separation, thus engineers relied on their experience and
engineering judgment to decide on these values. The Egyptian Code for Loads published in
1993 specified the minimum separation distance between two buildings as the larger of 2.0
cm or the maximum value of the horizontal sway of each of the two buildings (ECP-1993).
When the new code was issued in 2003, it has been scrutinized for the drift equation that
calculates the maximum inelastic displacement using a DAF equals to FRF, claiming its over-
conservatism (personal communication). The main concern was about the extremely large
building separation it resulted in, which is not acceptable from both a technical view because
of difficulty in using large expansion joints and an economical view because of limitations of
land use. In the draft code (2008), the ratio of DAF to FRF is specified as 0.7; however there
are still doubts about its overestimation of actual displacement. Some additional provisions
were added to shrink the estimate of separation of buildings when they have same floor
elevations by another 0.7 factor, and to get away from the whole calculation by providing
impact walls and fixing a separation of 4cm, or by designing the adjacent buildings to pound.
However, with the emerging construction of residential compounds of varying building
models and heights and due to the simplicity of the equivalent static method of analysis, the
use of the drift equation will be inevitable. Also if there is a problem with the estimate of the
maximum displacement as an absolute value, it will affect not only the building separation
size but also stability of the structural and nonstructural components as discussed before.
The inaccuracy of drift provisions was also pointed out for other seismic codes. In some
recent studies (Freeman and Searer, 2000) it was proved that the drift provisions in the 1997
UBC are extremely complex and overconservative. It is argued that even for inter-storey
drift, overconservatism results in difficulty in conforming to the code requirements because it
becomes too difficult to properly design and detail nonstructural elements such as cladding,
windows, and stairs, for unnecessarily large inter-story drifts. Also the requirement of large
separation material and sealant joints leads to compromises in the quality of construction that
can negate the purpose of the code and can interfere with the performance of waterproofing
of structures.
Also unlike strength design, conservatism in drift calculation can violate the safety as
well as the philosophy of seismic design of energy dissipation by ductile behavior where
7
Chapter 1
large drift estimates results in upgrading the cross section of members which increases their
stiffness. Increasing stiffness of members usually lead to an almost elastic behavior in which
any overloading can lead to the undesired sudden brittle failure modes.
Finally by comparing the value of DAF in different seismic codes, uncertainty of its
accuracy is even more exposed. It is well agreed that FRF values have their uncertainty
themselves and a comparison of FRF values across different codes is not very fair because
codes differ in the design philosophies, and safety and load factors used on the final design
values. Therefore for this reason and recognizing that DAF and FRF are interrelated, it would
be more rational to compare the ratio of DAF to FRF used in different codes (Maarouf et al.,
1996). A survey of seismic codes that included the 2008 Egyptian Code for Design and Loads
(ECP-201,2008), the 2004 Eurocode-8, the 1994 and 1997 Uniform Building Code (UBC
1994, and UBC 1997), the 1995 and 2005 National Building Code of Canada (NBCC, 2005),
the 2006 International Building Code of USA (IBC, 2006), and BCJ of Japan, shows that
different codes assign different values for this DAF. The results of this comparison are shown
in Table 1.1.
Table 1.1 Comparison between FRF and DAF used in different seismic codes
Building Code FRF DAF DAF/FRF
ECP-201(2008) R 0.7 R 0.7
Eurocode 8 (2004) q q
d
1 (unless otherwise specified)
UBC (1994)
UBC (1997)
R
W
R
3R
W
/8
0.7 R
0.375
0.7
NBCC (1995)
NBCC (2005)
R/U
R
d
R
o
R
R
d
R
o
U = 0.7
1
ASCE 7-05, IBC (2006) R Cd 0.5 1 (depending on structural system
and material)
BCJ 1/Ds Ds 1
Although this comparison is still not very objective because different codes use different
design seismic forces, return periods and acceptance criteria of drift checks, still the
considerable variation among DAF values with the absence of any technical justification
8
Chapter 1
incited the need for reliable calibration of the DAF used in the current Egyptian code for
seismic design of buildings.
1.3 RESEARCH OBJECTIVES
The main objectives of this research are:
(1) To evaluate appropriate DAF values to be used for seismic design of single and
multistory ordinary moment resisting RC frames in Egypt, by comparing the inelastic
displacement demands estimated by the current design procedures (equivalent static
lateral load method) according to ECP-201(2008) with the displacement demands
resulting from nonlinear dynamic finite element analysis, under strong ground motions
in accord with current design motion scenarios, as the closest representative of actual
behavior.
(2) To compare values of proposed DAFs for use across different seismic zones, namely
zone 1, zone 3 and zone 5B, and across different number of stories of buildings
(3) To calibrate the results achieved for zone 3 by modifying the earthquake input to reflect
the seismicity of the areas covered by this zone, using a representative ensemble of
natural records.
(4) To propose some adjustments to seismic design approaches for calculation of maximum
drift that bridge the gap between the calculated displacement using equivalent static
method of analysis and that obtained by nonlinear time-history analysis, by suggesting
disregarding code stipulated upper bound on the calculated fundamental period and
lower bound on the spectral acceleration, and verifying the resulting overconservatism
from their application.
(5) To investigate the sensitivity of the results to the following factors:
a. Building modeling assumption:
i. Reinforcing bars material constitutive model
9
Chapter 1
ii. Mean versus characteristic material properties
iii. Damping model
b.Earthquake Characteristics:
i. Duration
ii. Frequency Content
c. Irregularity represented by soft-first story
d.Effect of trading strength and stiffness in the design decision.
(6) To investigate the approach of different codes in incorporating the Importance factor for
buildings in evaluation of drift demands and in seismic design generally.
1.4 SCOPE AND WORK PLAN
Reinforced concrete (RC) moment-resisting frame buildings are chosen for study
because they represent the most common form of construction in Egypt, and their design is
likely to be governed by drift rather than strength. The number of stories were chosen as 1, 4,
7 and 10 to represent a variety of buildings commonly constructed in Egypt that can still
conform to the criteria of the code for application of the equivalent static load method. The
buildings are assumed to be for office use as typical for regular structures, and to be located
on soft soils. A range of design PGAs and seismic zones is included that covers the most
minor, most major and most menacing earthquake events.
Three-dimensional models of the buildings are created and analyzed for gravity load.
Then they are subjected to equivalent lateral load according to the code design spectrum of
zones 1, 3 and 5B, and designed accordingly. The resulting structures designed and detailed
to withstand seismic loads are used to model a two-dimensional moment resisting frame,
for drift analysis. Displacement analysis is performed by applying three analysis methods
namely: equivalent static load analysis, linear time-history analysis, and nonlinear time
history analysis. Seven artificial ground motion records are generated to match the design
spectrum for use in time-history analysis. The results of the three methods are compared as
applicable.
10
Chapter 1
For refinement of the model, the models located in Zone 3, which is the zone of highest
seismic risk, are subjected to real earthquake records, meticulously chosen to match the
design spectrum as well as represent the seismicity of the areas covered by this zone.
Moreover, all through the analysis process, a set of parameters is identified that could affect
the response obtained and the sensitivity of the results to these parameters is studied.
1.5 ORGANIZATION OF THE THESIS
This research comprises five chapters:
Chapter (1): briefly defines drift, and describes its effect on structures and the importance of
its control. The problem of the available methods for estimating drift and the inconsistency of
the proposed factors in different seismic codes is briefly outlined that signifies the need for
the current research, leading to the objectives of the work with definition of its scope and
work plan.
Chapter (2): presents a detailed appraisal of the factors and definitions related to the
characteristics of earthquake ground motions and the characteristics of structural responses to
ground motions, as well as the philosophy and criteria of seismic design codes in representing
earthquake action and the types of seismic analyses procedures. Also it explains the different
methods for estimating drift of buildings while conducting a detailed categorized review of
the available literature on similar studies.
Chapter (3): discusses in detail the methodology undertaken to achieve the defined
objectives. It explains the prototype models employed, the case study scenarios chosen, the
stages of analysis with the corresponding output and post-processing of results, the different
analyses performed with their associated assumptions, the modeling methods of members and
earthquakes, the software used, the assessment criteria for evaluation of DAF, and the
parameters identified for sensitivity study.
11
Chapter 1
Chapter (4): illustrates and discusses the results of displacement analysis of the 2-d frame
including assessment of the DAF factor proposed in the code, adjustment of the code
imposed-bounds for drift analysis, calibration of results against a refined model that takes
seismicity into consideration, and evaluation of the effect of the set of parameters studied.
Chapter (5): enumerates the main conclusions from this study identifying any limitations and
proposing recommendations for future work.
12
Chapter 2
CHAPTER 2
THEORETICAL BACKGROUND AND LITERATURE REVIEW
2.1 INTRODUCTION
In order to evaluate the approach of current seismic design practice in estimating
maximum inelastic displacement, there is a need to understand the concept upon which
seismic codes are based. Extensive research effort has been conducted in the past to
understand the characteristics of earthquake ground motions, and the characteristics of
structural responses to ground motions. The development of more reliable earthquake
resistant design methods depends on progress in the knowledge of seismic response of
structures based on past performances and observations of actual behavior during
earthquakes. This chapter provides a deep insight into the components of structure response
together with the effect of earthquake ground motion. Then it describes how the ensuing
design models these behavior and factors, with an understanding of the available analysis
tools and the limitations of modeling actual behavior. Finally methods for estimating drift of
buildings are presented while reviewing the available literature on similar studies.
2.2 BEHAVIOR OF BUILDINGS DURING EARTHQUAKES
According to Chen and Lui (2006), an earthquake is defined, from a structural point of
view, as a dynamic displacement of the ground supporting a structure, resulting in shaking of
the structure and consequent lateral and vertical forces. The ensuing forces are generated by
the structure rather than imposed on it like in other types of dynamic loading (Di Sarno and
Elnashai , 2008). This results in a complex interrelation between the characteristics of the
ground motion and those of the structure. Also unlike response to gravity loading which can
be analyzed on a member-by-member basis, response of buildings to earthquakes depends on
the global behavior of the whole system, as well as its correlation with the local responses of
13
Chapter 2
individual members and of sections up to the material level, which adds to the difficulty in
describing and modeling seismic behavior. The result is a complex interrelation of cause and
effect that makes factors influencing seismic behavior difficult to analyze individually
without incorporating into the others. Some of these influences are discussed below under
two broad factors: the nature of the ground motions and the characteristics of the building
itself.
2.2.1 The nature of earthquake ground motion
Most earthquakes result from sudden movement along the plane of faults within the
earth's crust releasing a great deal of energy in the form of seismic waves that travel through
the earth for great distances eventually losing most of their energy, and finally reaching the
earth's surface putting it in motion (Chen and Lui, 2006). This is called an earthquake ground
motion. When this earthquake ground motion is strong enough, and it occurs under a
building, the building vibrates.
2.2.1.1 Characterizing the Ground Motion
i. Ground motion time history
The earliest characterization of earthquakes ground motion was a qualitative
measurement of their effect, or intensity, at a certain point on earth where this measurement
differs for the same event from one location to another. Later, the magnitude of an earthquake
event became quantified by the amount of strain energy released at the source providing the
currently familiar Richter scale readings which are a unique value for each event. However,
for purposes of analyzing the response of structures to earthquake excitations, the time
variation of ground acceleration at a certain location is the most useful way to describe
ground motion during an earthquake. Numerical values of ground acceleration are recorded at
discrete closely spaced time intervals, resulting in a time-history record for the earthquake, or
accelerogram, which is output in an analog form on photographic film or as digital records
(Chen and Lui, 2006). Figure 2.1 shows the acceleration time-history of Al-Aqaba earthquake
that was felt in Egypt in 1995. The instrument used for recording strong-motion, called
14
Chapter 2
accelerograph, consists of three mutually perpendicular accelerometers, two measuring the
horizontal components of motion and the third measuring the vertical component.
Measurement of time variation of velocity and displacement are also sometimes deployed but
not widely used, and they are usually determined by integration of the time-history
acceleration record. Accelerograms describing past earthquakes are usually corrected for
instrumental and digitization errors, and filtered for short-period and long-period errors, and
compiled in databases for use in research and design.
Figure 2.1 Al-Aqaba (1995) earthquake ground acceleration: and corresponding 2%, 5%, 10% and
20% response spectra, as recorded on stiff soil in Eilat station (ESD)
ii. Response Spectrum
Another more general and convenient way of representing an earthquake and its effects
on structures, and which avoids the complex random nature of a ground motion record, is the
response spectrum, which is a fundamental concept in earthquake engineering. Almost all
strong ground motions that are recorded from past earthquakes have their corresponding
response spectra computed and published. A response spectrum is defined as a plot of the
maximum value of a response quantity (displacement, velocity or acceleration) of a complete
range or spectrum of linear single-degree-of-freedom (SDOF) to a known ground motion as a
function of their natural period and for a fixed damping ratio. Many plots are developed to
cover several damping values. The result is a practical means of characterizing a certain
15
Chapter 2
earthquake which summarizes the peak response which is most important from a structural
point of view of all possible structures idealized as linear SDOF systems to a particular
component of the ground motion (Chopra, 2005). Each point on the response spectrum curve
is defined by taking the maximum response from the structural response time history of a
SDOF system generated by dynamic analysis of the system when subjected to a given motion
time history as shown in Figure 2.2, and the process is repeated for structures of varying
periods to obtain the whole response spectrum. The shape of the response spectrum curve
represents the properties of the earthquake at a specific site and do not depend on the
properties of the structural system (Chopra, 2005). A typical response spectrum for AL-
Aqaba (1995) earthquake is shown in Figure 2.1
Figure 2.2 Graphical description of computation of an earthquake response spectrum
Usually the displacement response spectrum is constructed and two related response
spectra, pseudo-velocity and pseudo-acceleration are calculated by multiplying the
displacement spectral ordinates by (2/T) and (2/T)
2
respectively, where T is the period read
on the abscissa. These are approximations of the true velocity and acceleration spectra that
can be constructed by dynamic analysis similar to the displacement spectrum. Earthquake
response spectra can be plotted on a four-way logarithmic scale to display displacement,
16
Chapter 2
pseudo-velocity and pseudo-acceleration all on one plot, called tripartite spectrum. The peak
strain energy stored in the system during an earthquake is related to the pseudo-velocity
values, and the inertia forces developed in members are proportional to the pseudo-
acceleration (Chopra, 2005). Therefore, because of their physical interpretation pseudo-
acceleration spectra have the most practical application for design purposes (Bertero and
Bozorgnia, 2004). Pseudo-acceleration is total acceleration not relative values like
displacement and pseudo-velocity. For very stiff structures (with near zero vibration periods),
the relative acceleration will be nearly zero, and thus pseudo-acceleration, which is the total
acceleration, will be equal to the peak ground acceleration. Pseudo-acceleration spectrum is
almost the same as the true total acceleration response spectrum when there is no damping;
however differences appear with increasing damping values.
2.2.1.2 Ground motion characteristics affecting structural response
It can be observed from the accelerogram shown on Figure 2.1 that ground motions due
to an earthquake are very complex and vary with time in a highly jagged manner, as waves of
different frequencies and amplitude interact with one another. The main physical
characteristics that describe a ground motion are: frequency content, peak amplitude
(acceleration, velocity and displacement) and duration of the shaking.
i. Frequency content
The time in seconds required to make one complete cycle of vibration is called the period
of vibration. Frequency is the reciprocal of the period where it denotes the number of
complete cycles made by the wave per second. Ground motion at a particular site is a
complex superposition of different vibration frequencies, with certain frequencies usually
predominating. The distribution of frequencies in a ground motion is known as its frequency
content. A vibrating building also has frequency content, though one particular frequency
dominates the building's vibrations this is called the buildings natural or fundamental
frequency. When the frequency contents of the ground motion are close to the building's
fundamental frequency, the building and the ground motion are said to be in resonance with
one another, and the building's response is highly amplified (Taranath, 2005). The frequency
17
Chapter 2
content of ground shaking can be defined either by transforming the ground motion time
history from the time domain to the frequency domain through its Fourier transform (Clough
and Penzien, 1993), or by representing the ground shaking as an acceleration response
spectrum relating ground acceleration to frequency which is the more commonly used
method in seismic analysis (Chopra, 2005). Single-value indicators of frequency content are
used, like the characteristic energy period (Tce), predominant velocity period (Tg),
characteristic period (Tc), predominant energy period (Tes) and long period (Tlp). The
definitions of these indicators are beyond the scope of this study and the reader is referred to
Miranda (1991, 1993), Shimazaki and Sozen (1984), Vidic et al (1994) and Mukesh et al.
(2010, in press), for further information.
ii. Peak amplitude
This represents the largest value of a certain response of the ground like acceleration,
velocity or displacement. Peak ground acceleration (PGA) is currently the most important
factor from an engineering point of view because according to Newton's Second Law the
resulting inertial forces in a structure are directly proportional to the acceleration. It is
measured as the maximum absolute amplitude on a recorded or synthetic acceleration time
history. However, in many cases, peak acceleration corresponds to high frequencies which
are far from the natural frequencies of most structures that fall in the low-to-intermediate
range, thus cannot initiate resonance (Werner, 1991). And also peak acceleration alone does
not give a proper indication of damage without considering the duration of shaking because a
longer smaller acceleration than the peak acceleration can result in more deformation in a
structure (Singh, 1995). Anderson and Bertero (1987), Uang and Bertero (1988), and Bertero
et al. (1991) have indicated that earthquake ground motion characteristics such as frequency
content, duration, velocity, and displacement can affect the response of structures much more
intensely than the peak ground acceleration, especially in the inelastic range. However, PGA
is till-present the key aspect of definition of seismic hazard in seismic design codes including
the ECP-201 (2008), where it represents the first point on the elastic design response
spectrum as will be discussed in Section 2.3.2.
18
Chapter 2
iii. Duration of strong shaking
Duration of strong motion has a profound effect on the level of damage and destruction.
Usually the longer the duration is, the less acceleration the building can survive, and the
longer the duration is, the more the damage for the same acceleration. This is especially
important for structures designed to be earthquake-resistant because they are designed to be
ductile, and ductile response depends on the history of deformation and the number of cycles
of inelastic deformation, therefore the duration of ground motion may have a considerable
influence on the amount of energy dissipation after cycles of deformation (OConnor and
Ellingwood, 1992). Nevertheless, duration of ground motion is not currently directly
considered in common seismic design practice except through time history analysis
(Taranath, 2005).
Different procedures for computing the duration of strong motion are available in the
literature. The bracketed duration proposed by Page et al. (1972) and Bolt (1969), is the
method still used in studying the elastic and inelastic behavior of structures. This is the time
interval between the first and last acceleration peaks greater than a specified value usually
0.005 g on an accelerogram (Naeim, 1989).
2.2.1.3 Factors affecting ground motion characteristics at a particular site
The main factors that affect the ground motion characteristics at a certain location are:
source effects which are the earthquake magnitude and the source conditions and mechanism
(fault type, rupture process and stress conditions), travel path effects which are the distance
from the source and variations in geology and propagation velocity along the travel path,
local site effects which are the local soil conditions and soil-structure interaction effects.
i. Source Effects
Ground motion at a particular site is to a large extent determined by the magnitude of the
earthquake and characteristics of the source mechanism. As expected the larger the
earthquake magnitude the larger the peak ground accelerations, velocities and displacements
19
Chapter 2
(Naeim, 1989). Earthquake magnitude also increases the duration of strong motion as
concluded from several studies by Housner (1976), Donovan (1973) and Page et al (1972).
However it should be noted that this effect of magnitude on duration is combined with other
effects due to distance from epicenter and soil type. As for the effect of earthquake magnitude
on frequency content, studies show that increasing magnitude leads to higher spectral
amplitudes on a response spectra with especially higher increase at long periods (Anderson
and Quass, 1988). Source conditions affect the magnitude of ground motion and they include
the depth of the source, size of the area of rupture, the amount and distribution of
displacement (slip) on the fault plane, rise time which is the time for the slip to complete at a
given point on the fault plane, and the faulting type (Chen and Lui, 2006). Subduction zone
earthquakes result in large and relatively deep earthquakes having response spectra lower in
the long period range than the response spectra for shallow crustal earthquakes (Youngs et
al., 1993). Directivity effects and Near-source effects of pulses and flings also affect ground
motion direction and characteristics (Naeim, 1989). This results in higher spectral values
especially in the direction normal to the fault strike (Somerville et al., 1997).
ii. Travel path effects
As distance from the source mechanism increases, the amplitude of vibration of ground
motion is reduced. During an earthquake, waves emitted from a source differ in their shape
and velocity and their capacity of propagation through different media. Therefore that nature
of the geological formations between the source of the earthquake and the building impinge
on the type of waves arriving at a building. On their way through the earth, the waves are
modified by the soil until reaching the surface in a different waveform (Naeim, 1989). The
reduction of amplitude, called attenuation, occurs at a faster rate for higher frequency (shorter
period) components than for lower-frequency (long period) components (Taranath, 2005).
Therefore a generalized effect of increasing source-to-site distance on the response spectrum
shape is a lower curve in the high-frequency range and a higher curve in the long-period
range (Silva and Green, 1989). Many attenuation relationships for amplitude with distance
have been presented by many investigators, some magnitude-dependent (Housner, 1965;
Donovan, 1973, Seed and Idriss, 1982 and Mualchin and Jones, 1992) and others independent
on the earthquake magnitude (Naeim, 1989). The effect of magnitude is much greater than
the effect of epicenter distance within 50km of an earthquake source (EM, 1999). Also the
20
Chapter 2
duration of motion decreases with an increase in distance from the epicenter as concluded by
Page et al (1972), Trifunac and Brady (1975), and Chang and Krinitzsky(1977), however, the
strong shaking part of the accelerogram has a longer duration and this effect is generally
small for source-to-site distance of less than 50km (EM, 1999).
iii. Local site conditions
The type of soil layers in the vicinity of a building site affects the attenuation of ground
motion. Many researchers such as Seed and Idriss (1982) provide attenuation relationships
for different types of soil. They conclude that assuming a fixed distance from the source,
peak accelerations on rock is greater than those on soft soils. At the same time, harder soils
and bedrock are able to transmit short-period vibrations (caused by near field earthquakes)
while filtering out longer-period vibrations (caused by distant earthquakes), whereas softer
soils will transmit longer-period vibrations (Taranath, 2005). But since shorter-period
vibrations are attenuated at a higher rate than longer-period vibrations, structures built on
harder soil perform much better than those on soft soil. Soft soils are known to amplify the
ground motion; however this amplification depends on the ground motion acceleration
amplitudes, where for acceleration motion higher than 0.4g, nonlinear behavior of softer soils
leads to more damping and de-amplification of the high-frequency response, and less
amplification in the longer period range (EM, 1999). Soil conditions also affect the duration
of strong motion, where for a given earthquake magnitude, the duration of strong motion in
softer soils is almost double that in rock (Chang and Krinitzsky, 1977).
Moreover from a structural response point-of-view, the topography of a building site can
also affect the structures response and damage potential. Buildings built on steep ground,
strip-shaped hill ridges, or separated hill ridges are more damaged than those built on an even
ground with open space around. Also, soil-structure interaction affects the response of the
structure and its dependence on the soil conditions. For soft soils, the mass of the building has
a more negative effect on the building response if the building is stiff and heavy, while the
density and flexibility of the soil is the governing factor if the building is light and slender.
Soil-structure interaction in soft soils can also significantly reduce the amplitude of vibration
(Di Sarno and Elnashai, 2008).
21
Chapter 2
2.2.2 Response of Structures to Ground Motion
2.2.2.1 Characteristics of Building vibration in response to earthquake excitation
Earthquakes cause buildings to move in a dynamic and reversed manner which makes
the behavior of the building like any vibration problem, with an added complexity due to the
erratic nature of seismic waves. The three fundamental characteristics of building vibration
are: amplitude, mode shapes and frequencies and damping.
i. Amplitude of vibration
The amplitude is the maximum value of vibration at a given location on a structure. The
amplitude of any vibration depends on the force initiating it, however, in the case of
earthquakes, and as previously discussed, forces are generated by the structure itself. When a
building is subjected to ground motion beneath it, only the base of the building moves with
the ground and the rest of the superstructure resist this motion and tend to remain at its
original position. Thus the building appears as if pushed in a direction opposite to the
direction of ground motion. The unseen force responsible for this push is called the Inertia
Force. In accordance with DAlemberts principle, this apparent force is equal to the mass
multiplied by the acceleration with a negative sign to account for the opposite direction of the
acceleration. This is equivalent to Newtons law concerning actual physical forces.
It follows that the amplitude of vibration depends on the mass of the structure as well as
the ground acceleration it is subjected to. The mass of a structure is generally assumed a
constant throughout the response. In the case of designing a new building, reducing the mass
of the structure improves its overall seismic behavior because it leads to reduced forces
(Taranath, 2005). On the other hand, the amount of acceleration is variable for different
structures and different ground motions. Acceleration depends on the buildings natural
frequency as compared to the frequency content of the input ground motion, which depends
on other properties of the system as will be discussed later. This is one example of the
interrelationship between structure response parameters and ground motion parameters. The
amount of acceleration reaching the building depends also on the soil-structure interaction as
described in Section 2.2.1.2 (iii). Modern methods of base isolation, which separates the
22
Chapter 2
movement of the base of the building from its superstructure, reduce the amount of
acceleration the building feels and thus reduces the amplitude of vibration.
ii. Vibration Mode shapes and frequencies
Every building has a number of ways or modes in which it can vibrate naturally. A mode
shape describes the expected curvature (or displacement) of a surface vibrating at a particular
mode. Typically, a building will vibrate in a combination of its natural frequencies and mode
shapes at once, however the lowest natural frequency defined as the fundamental frequency is
usually the predominant frequency as measured by the percentage of building weight
vibrating at it. The distorted shape it takes up when vibrating at this frequency is called the
fundamental mode shape. Figure 2.3 shows the fundamental mode shape as well as the
second and third mode shapes of vibration of an example building. In an earthquake, as
argued before, if the ground motion frequency content coincides with any one of the
building's natural frequencies, then the building acceleration will dramatically be amplified
due to resonance.
Figure 2.3 Typical first three mode shapes of a building
The frequencies at which vibration naturally occurs, and the modal shapes which the
system takes, are properties of the system. The fundamental or natural period, which is the
inverse of the fundamental frequency, is a more commonly used term for buildings because it
is easier to visualize for structures and it depends on (Chopra, 2005):
a. stiffness: for the same mass, the higher the stiffness is, the shorter is the natural period
b. Mass: the heavier of two structures with the same stiffness has a longer natural period.
23
Chapter 2
c. Boundary conditions of the columns: the natural period of a frame with hinged-base
columns is almost double that of a frame with clamped-base columns, because it has
lower stiffness.
The natural period also depends on the damping characteristics which depend on many
factors as will be discussed shortly (Taranath, 2005). And all these factors of mass, stiffness
and damping depend on the material used in construction and the structural system employed
(Taranath, 2005).
It should be noted that a smaller period corresponds to a higher response on a response
spectrum and therefore larger base shear calculations as will be discussed in Chapter 3.
Thats why seismic codes usually put an upper bound limit for the fundamental period to be
used in design.
iii. Damping of Vibration
For all vibrating objects, including buildings, the amplitude of vibration tends to decay
with time until the vibration stops. The process that diminishes the amplitude of free
vibration until bringing the object back to its original static state is called damping (Chopra,
2005). Without damping, an object once set in motion, will vibrate indefinitely.
Damping is essentially dissipation of the energy of the vibrating systems by several
physical processes that may happen at the same time. These processes include:
External mechanisms: viscosity of the air or water surrounding the building (results in
insignificant damping) (Taranath, 2005); radiation of seismic waves away from
foundations (Di Sarno and Elnashai, 2008); radiation of sound waves.
Internal mechanisms:
- Viscous damping: on the material-level this results from thermal effects due to repeated
elastic straining of the material and internal friction when a solid is deformed (Chopra,
2005). It depends on the viscosity or ductility of the material. On the structure-level, this
is represented by the area under the force-deformation curve and is proportional to the
velocity of vibration and changes in proportion to the natural frequency of the structures
vibration (Taranath, 2005).
24
Chapter 2
- Friction damping: friction at connections and support joints, and friction between the
nonstructural elements and the structure such as partitions, ceilings and in-fills (Chopra,
2005). This kind of friction, called Coulomb damping, usually doesnt depend on the
velocity or amount of displacement therefore it is taken as a constant depending on the
material and type of construction (Taranath, 2005).
Unlike mass and stiffness properties, damping properties and mechanisms cannot be
identified precisely and separately in an actual structure; therefore a mathematically
convenient approach is to represent damping in a structure by an idealized coefficient, called
the damping coefficient (c) of an equivalent linear viscous damper (dashpot), which is
evaluated based on vibration experiments on actual structures and is chosen to dissipate
vibrational energy equal to the total energy dissipated per cycle of vibration of the structure,
by the various damping mechanisms lumped altogether (Di Sarno and Elnashai , 2008). And
since every object or building have a different degree of intrinsic damping, a critical damping
is defined for every object as the value of minimum damping required to prevent oscillation
at all. It depends on mass and stiffness of the structure. In order to use a dimensionless
property when describing damping, damping is usually presented in earthquake engineering
as a ratio or fraction of critical damping, called the damping ratio , which is a property of the
system material and independent of its mass and stiffness (Chopra, 2005). Damping ratios
commonly used in practice range between 1 and 10% of critical damping (Taranath, 2005).
Damping values depends on the construction materials, vibration amplitude, fundamental
period and mode shapes, type of connections and the building configuration (Di Sarno and
Elnashai, 2008).
It is worth mentioning that the damping ratio is intended to model energy dissipation
within the linear behavior of the structure, and the value of the damping coefficient is chosen
from experiments to correspond to the most critical deformation which is associated with the
linear elastic limit of the structure, ignoring any nonlinearity of the damping with
deformation amplitude before the elastic limit (Chopra, 2005). At larger deformations,
beyond the elastic limit, a new form of energy dissipation occurs due to inelastic behavior of
the whole structure, which is not suitable for the equivalent viscous damping idealization. It
is accounted for by methods recognizing the inelastic force-deformation relationship as
discussed in Section 2.2.2.2. The force-deformation curves are obtained from testing on
25
Chapter 2
structures at slow rates of deformation in order to rule out any energy dissipation resulting
from rate-dependent effects which are already accounted for by the viscous damper (Chopra,
2005). Since the earthquake loading is cyclic, yielding is repeated several times in a process
called hysteresis. More about hysteretic damping is discussed in Section 2.2.2.3.The amount
of hysteretic damping is much more than the intrinsic damping of the structure and is thus
desirable in strong earthquakes, provided that measures are taken to control damage due to
permanent deformation resulting from inelastic behavior.
In modern earthquake engineering, other advanced systems are available that employ
adding energy dissipation devices to buildings to enhance artificially its intrinsic damping
and thus improve its earthquake performance without relying on any inelastic behavior and
thus avoiding suffering any damage. Many types of supplemental damping devises have been
developed for this sake that dissipate energy by friction, or yielding of metals, or controlled
shearing of visco-elastic solids, or forced movement of viscous fluids (Chen and Lui, 2006).
They are usually used for seismic retrofitting of structures or to reduce drift in new structures
(Soong and Spencer, 2002).
2.2.2.2 Describing structural response to earthquakes
Structural response to earthquakes is usually evaluated in two ways that complement
each other. The first is the structural Response Time-History (RTH), and is the result of a
demand study, where the response of a structure to a particular earthquake on a structure is
evaluated (at a particular level of loading), and the second one is the structure load-
deformation relationship, which is the result of a supply study, where the whole relationship
of force and displacement of the structure is described (at all levels of loading).
1
1
Interestingly, the relatively recent method of incremental dynamic analysis - in which the structures response
time-history is analyzed at increasing levels of the same ground motion acceleration provide a means of
performing a demand study of the capacity of the structure and creating a load-deformation curve for the
structure when subjected to this particular earthquake.
26
Chapter 2
i. Structural Response Time-History (RTH)
A structure can be idealized as a SDOF system, and the ground motion time history
applied to it as a time-varying load. By applying dynamic equilibrium at any point in time,
the sum of all resisting forces is equated to the applied load. The resisting forces are inertial
forces, damping forces and elastic forces. Therefore,
F(t) fI(t) fD(t) fS(t) = 0................................................................................................(2.1)
Where,
F(t) is applied time varying earthquake load
fI(t) is inertial force
fD(t) is damping force
fS(t) is elastic force
Based on experimental results, relationships are available in the literature that relates each of
these resisting forces to the three motion parameters relative to the fixed base of the structure:
displacement ur(t), velocity r(t) and acceleration r(t). These relationships are:
fI(t)= mr(t)..........................................................................................................................(2.2)
fD(t)= cr(t)..........................................................................................................................(2.3)
fS(t) = kur(t)..........................................................................................................................(2.4)
Where,
m is the total mass
c is the damping coefficient
k is the stiffness coefficient
And since earthquake forces are actually generated inertia force not applied load, therefore:
F(t) = - mg(t).......................................................................................................................(2.5)
Substituting equations (2.2) till (2.5) into (2.1)
27
Chapter 2
mr(t)+ cr(t) + kur(t) = - mg(t).........................................................................................(2.6)
Where, g(t) is the ground acceleration time history
Substituting c/m = 2 and k/m = 2 into equation (2.6) and dividing by m
r(t)+ 2 r(t) + 2ur(t) = - g(t).......................................................................................(2.7)
The response history is generated by integrating the solutions to this differential equation
(2.7) to cover the whole time-history of ground motion, applying numerical time-steeping
methods like central difference method or Newmarks method. From the equation of motion,
it can be concluded that the response history is defined by the damping ratio and natural
frequency (which are the key characteristics of building vibration as explained in Section
2.2.2.1) and the ground acceleration record. The response history is the input employed in
constructing the response spectrum.
Since structures are expected to crack and yield during an earthquake, inelastic behavior
should be accounted for. In this case the force corresponding to deformation is not single
valued, therefore it is replaced by a force-deformation relationship fs(u, ) in equation (2.6),
resulting in the following equation of motion for inelastic systems:
mr(t)+ cr(t) + fs(u,) = - mg(t)........................................................................................(2.8)
The force-deformation relationship of the resisting force fs(u, ) can be idealized for an
elastoplastic system and the equation of motion solved numerically using time-stepping
methods to achieve an inelastic response time-history. An example of elastic and in-elastic
response history analysis is shown in Figure 2.4.
28
Chapter 2
Figure 2.4 Elastic and inelastic response time-history results (Bazzuro et al., 2004)
ii. Load-deformation relationship
The actual behavior of all objects under loading whether a material, element, or a
complete system can be described by a load-deflection curve. In the case of buildings
subjected to earthquake loading, this load deflection curve relates the total base shear to the
displacement at the top of the building. For materials the behavior is described by stress
versus strain (axial or shear) relationships, and for sections and connections, load versus
deflection curves, and moment versus curvature or rotation curves are used respectively. The
following discussion concerns the system force-displacement curve of a structure under
earthquake excitation. It is either established by laboratory testing where the relationship is
defined as an idealized version of the experimental data or by numerical tools that push a
building to failure known as a pushover analysis (nonlinear static structural analysis) in
which a stress-strain law is assumed for the material, and the analysis trails the beginning and
spreading of yielding at critical locations to get the initial loading curve, and the unloading
and reloading curves are either computed similarly or generated from the initial loading curve
using existing hypothesis (Chopra, 2005) .
The choice of the base shear is based on the concept that since the seismic forces are
inertia forces, the total resisting forces affecting the building progressively increase as we
approach the bottom of the structure, because more mass is being carried, reaching the
maximum at foundation level. The maximum displacement, on the other hand, occurs at the
29
Chapter 2
top of the building because it is the base of the building that moves with the earthquake, and
thus the top is the point most lagging behind, and suffering maximum displacement. This is
actually only true if the structure is assumed to vibrate in its fundamental mode of vibration,
while for higher modes, the load deflection curve should be based on the total base shear
versus maximum or average inter-story drift.
Figure 2.5 Typical base shear-displacement curve for a building subjected to horizontal loads
The initial loading curve created under monotonic loading provides an envelope for the
complete cyclic load deformation relationship. This curve consists of four stages as shown in
Figure 2.5. In Stage (1) un-cracked linear stage: the structure is responding elastically, and if
the force is removed the structure will return to its original shape without any permanent
deformation and thus no damage. Stage (2) cracked-linear stage: starts at the first sign of
damage; since there is damage, the stiffness of the structure changes, and it is deforming
quickly for a smaller increase in force. This stage is governed by the cracked stiffness. Stage
(3) cracked-nonlinear stage: starts with the first sign of yield in a system as defined by
formation of the first full plastic hinge at the most critical hinging region of the structure
(usually close to connections). If the force is removed, there will be some residual permanent
deformation and this stage is governed by the post-yield stiffness. Due to the ductility and
30
Chapter 2
strain-hardening of the hinging region and the redundancy that is built into the system,
different critical regions attract more moments and yield in their turn forming a series of
plastic hinges in the structure, allowing even more capacity to sustain load, finally leading to
a complete yielding mechanism at the strength level Vy. When the first hinges to form reach
their rotational capacity, they begin to lose strength, and the curve begins to flatten out.
Stage (4) started when the structures capacity to resist deformation is declining which is
termed the softening stage. Near the end of the curve where the slope is declining, any
increase in force will lead to failure.
2.2.2.3 Fundamental Parameters controlling Structures Seismic Response
The basic building blocks for understanding and expecting response of structures when
subjected to earthquakes are: mass, stiffness, strength and ductility. They are not period and
damping because these are actually a consequence of the fundamental parameters as
discussed before. For example, the period of vibration depends on the mass, stiffness and also
the strength in the inelastic range; and the major source of damping in most common
structures not supplemented by modern damping devices is the energy dissipated by inelastic
deformation which depends on ductility (Di Sarno and Elnashai, 2008). Since mass is a fixed
term for each structure and can be readily estimated based on the effective weight, focus is
made on the other three parameters: stiffness, strength and ductility. These parameters are
control values that the designer chooses and consequently dictate the behavior of the
structure. Stiffness, strength and ductility of a structural system are affected by the same
properties along a hierarchical relationship of material, Section, member, connections up till
the whole system. Therefore they depend on the mechanical properties of the material of
construction, geometrical properties of members and lateral-load resisting system as well as
quality of construction. The regularity and uniform distribution of these properties in the
structure also affect the structure response.
i. Stiffness
Stiffness represents the ability of the component to resist deformations when subjected to
forces. This can be considered the opposite of flexibility. It is expressed as the ratio of
31
Chapter 2
deformation to the associated load level. For a building it is represented by the ratio of top
displacement (or inter-storey drift if irregular building) to the total base shear in the elastic
range, as shown in Figure 2.6. This initial slope K0 is called the elastic stiffness of the
structure. In an RC building which is expected to behave in-elastically by cracking of
concrete and yielding of steel bars, it is difficult to define a fixed value for the stiffness.
Secant stiffness, or cracked stiffness which is the slope Ks corresponding to a certain load
level taken at the yield point is sometimes used to represent the stiffness of RC structures to
avoid assuming a larger stiffness than actual by using K0 for a cracked structure. Another
mathematical representation of stiffness is the tangent stiffness Kt, also illustrated in Figure
2.6, which is used to indicate the rate of stiffness change in the post-yield region, and is
usually used in incremental studies like time history analysis for faster convergence than
secant stiffness (Di Sarno and Elnashai, 2008).
Figure 2.6 Definition of stiffness (Di Sarno and Elnashai, 2008)
The stiffness of a structure depends on the type of lateral load-resisting system chosen,
the stiffness of its connections, members and materials of construction. For example,
structural walls are stiffer than frames. Soil structure interaction can also reduce the stiffness
of the superstructure (e.g. Mylonakis and Gazetas, 2000). A group of researchers studied
experimentally the influence of changing connection stiffness on the stiffness of a two-storey
steel frame proving that a reduction of 50% and 60% in connection stiffness results in 20%
and 30% reduction in total building stiffness respectively (Elnashai et al., 1998). Member
stiffness depends on the geometrical properties of the member such as section dimensions,
and height and aspect ratio as well as stiffness of the material and boundary conditions.
Section stiffness in its turn is calculated based on the cross-sectional area A, the flexural and
32
Chapter 2
torsional moment of inertia I and J and these properties themselves depend on the loading and
boundary conditions for RC structures. Material stiffness is measured by the modulus of
elasticity, which is the ratio of stress to strain in the elastic range of material behavior
(Youngs moulus E and shear Modulus G). In the inelastic range, the stiffness is evaluated by
the tangent values to moduli E and G. Material stiffness is usually given by the relative
elasticity ratio which is E/, where the lower the value of E/ the stiffer the material. Concrete
is stiffer than metals and masonry is stiffer than concrete.
Stiffness affects the structures response in various ways. The higher the stiffness, the
shorter the natural period of a structure, which means the faster the vibration. Stiffness affects
deformability of the structure and hence its functionality. The lower the stiffness is, the less
the higher is the deformability and the worse is the serviceability of the structure. Stiffer
elements on the other side attract more seismic load. Therefore non-uniformity of stiffness
distribution in plan or elevation leads to damage concentration because stiffer elements that
attract seismic load reach their capacity earlier leading to failure. This non-uniform
distribution of stiffness also results in torsional effects. Moreover, the relative stiffness of
beams and columns affect the behavior of the system by either providing restraint or no
restraint for columns at the beam-column connection and increasing the bending moment in
beams (Di Sarno and Elnashai, 2008).
ii. Strength
Strength represents the ability of the component to sustain load. Therefore it is a force or
stress quantity corresponding to a defined damage state of deformation or strain. The damage
state can be either yielding (permanent deformation) or ultimate failure. The term capacity
means the capacity to endure deformation which is the same as the ability to sustain load.
Yield strength Vy corresponds to the load at which first yield or damage occurs, and ultimate
strength Vmax corresponds to the maximum deformation at failure. These definitions are
illustrated in Figure 2.7. Capacity can be expressed as axial, bending and shear resistances.
Under earthquake ground motion, the structure should possess enough strength to resist base
shear forces and overturning moments.
33
Chapter 2
Figure 2.7 Definition of Strength (Di Sarno and Elnashai, 2008)
The strength of a structural system depends on the strength of its members, connection
and the material of construction, but its not the sum of their strengths because they interact
in a complex way. Material strength can be quantified by testing resulting in the stress strain
curve. Material Strength is often referred as a ratio of weight. Section strength in RC
members depends on the dimensions of the section, the moment of inertia and the amount of
reinforcement.
There are many factors that results in change in the strength of the structure than the
initial estimated value. Uncertainties due to randomness in material properties and section
sizes and construction quality, affects the credibility of the estimate of strength for a
structure. At the material level, strain hardening and softening affects strength. At the section
level, tensile strength is much less than compressive strength, however, due to load reversal,
sections are subjected to tension and compression and columns are subjected to biaxial
moments, which affects the ability to sustain load. The interaction between shear and flexure
actions due to reversed loading significantly affects strength. As for connection properties,
loss of strength at the beam-to-column and column base connections due to complicated
geometric arrangement or congestion of steel bars leads to significant reduction in strength of
structure. Also loss of shear strength at the connection between slabs and columns or
structural walls can result in punching or tearing.
34
Chapter 2
iii. Ductility
Ductility is another factor that significantly affects the performance of a building during
an earthquake. Ductility reflects the ability of a component to undergo large inelastic
deformations before failure. While strength represents a force capacity, ductility is inelastic
deformation capacity. As illustrated in Figure 2.8, brittle systems reach failure after very
small inelastic deformations. Ductility concerns the behavior of a material, element or system
beyond the elastic range. Ductile behavior allows energy absorption, thus damping of the
vibration. Also due to the large inelastic deformations involved, failure in a ductile structure
is not sudden. Moreover, when ductile elements dissipate energy, a series of successive
plastic hinges are formed, because stiffness is reduced in these dissipative zones resulting in
force migration to other stiffer zones. This action redistribution because of successive
yielding results in prolonging the life of structure. Therefore, ductility is a property favored in
seismic design. Nevertheless, allowing ductile behavior entails controlling damage.
Figure 2.8 Definitions of ductile and brittle structures (Di Sarno and Elnashai, 2008)
Ductility is mathematically defined as the ratio between the maximum inelastic
deformation and the deformation at yield, as shown in Figure 2.8. Material ductility
is
defined as the ratio of maximum inelastic strain to strain at yield stress on a stress-strain
curve obtained by testing of the material. Section curvature ductility
is the ratio of
maximum plastic curvature to curvature at yield on a moment-curvature diagram obtained by
applying moment on a cross section. Member rotation ductility
is the global measure of ductility of the structure measured as the ratio of maximum
inelastic displacement to displacement at yield. The hierarchy of ductility levels in shown in
Figure 2.9. It should be noted that ductility decreases along the hierarchy from top to bottom,
for example, a material ductility of 5 or 6 results in system displacement ductility of 2 or 3.
Figure 2.9 Hierarchy of ductility in a structure (Di Sarno and Elnashai, 2008)
Global ductility of a system depends on the local rotational ductility and curvature
ductility which in their turn depend on ductility of the material of construction. Concrete is
fairly brittle and steel reinforcement is used to improve its ductility. Ductility of RC is highly
improved by confinement by transverse steel reinforcement which increases ultimate concrete
strains and thus increases curvature ductility. Using high strength steel increases the yield
strength of steel which comes in the denominator in the definition of ductility and therefore
36
Chapter 2
reducing curvature ductility. Similarly increasing ultimate strength to yield strength ratio of
steel enhances curvature ductility. Ductility of a member depends on the spreading of
plasticity in the critical region. The elongation of reinforcement at this region results in
additional deflection and rotation in the member. Rotational ductility of a joint depends on
joint dimension, amount of steel reinforcement, bond resistance, level of column axial loads
and presence of slab and transverse beams framing into the connection.
Modern seismic codes usually start with an assumed ductility as reflected partly through
the force reduction factor. However it should be noted that this presumed ductility is not
always achieved due to the following factors:
Strain rate effects that causes increase in strength
Reduction in capacity to absorb energy under cyclic loading
Over-strength leading to structures not to yield when they were intended to yield thus act
as if brittle
Unintended brittle failure of materials in the system.
2.2.2.4 Dynamic Hysteresis
For earthquake cyclic loading, the force-deformation relationship explained in Section
2.2.2.2 (ii) goes through a series of loading and unloading in reversed directions generating
loops called hysteresis loops. The amount of energy dissipation per cycle of vibration equals
the total area under the loop for this cycle. And because the deformation is inelastic, the
unloading and reloading branches do not coincide with the initial loading curve due to the
residual deformations in each cycle, leading to a different amount of energy dissipation. This
shows that the inertial forces developed in a structure is not single-valued and depends on the
history of deformation. A distinctive characteristic of dynamic hysteresis is that the hysteresis
loops are rather elliptical rather than pointed in shape (Chopra, 2005). The hysteresis
behavior of reinforced concrete members may exhibit some of the following features as
illustrated in Figure 2.10
37
Chapter 2
Figure 2.10 Hysteresis behavior of RC structures (Seoudy et al., 2000)
38
Chapter 2
i. Stiffness degradation
During the loading and unloading cycles, degradation of stiffness occurs as marked by
the gradual reduction in the slope of the force deformation hysteresis curve after each cycle.
Some significant stiffness degradation is unavoidable, because of the change in member cross
sections by damage due to yielding. Strain softening which is the reason for stiffness
degradation can be reduced by confining concrete. The main effect of stiffness deterioration
on the structure is elongation of the period of vibration in the inelastic region (Elnashai and
Mwafy, 2002). This results in a continuous change in the stiffness and period of structure and
in its turn continuous change in the amount and distribution of internal forces which could
lead to excessive deformations in some regions.
ii. Strength degradation
Strength degradation leads to lowering the response curve thus reducing the energy
dissipation capacity of the material, therefore significant loss of strength should be avoided.
The loss of strength is mainly due to loss of bond under repeated loading between concrete
and steel in RC structures (Di Sarno and Elnashai, 2008), and once it starts it is progressive.
The amount of strength deterioration depends on many factors including confinement of
concrete, the governing deformation mode, shear strength, loading history, and the level of
axial load.
iii. Pinching of Hysteresis loops
Hysteresis loops of reinforced concrete members generally show a marked change in
slope during reloading. This is because the plastic strain accumulated in the steel
reinforcement results in keeping the cracks open at the end of unloading, and when the load is
reversed, this old crack closes and a new direction cracks open, creating a gap of very small
stiffness because of the accumulated plastic strain in the reinforcement. When this
reinforcement plastic strain increases excessively, it delineates the crack closing and opening
during load reversal (Adebar and Gerin, 2004). Pinching results in slimmer and more pointed
loops, and thus leads to reduction in energy dissipation capacity. Flexural deformations
usually do not lead to pinching effect as large as that resulting from inclined shear cracks.
39
Chapter 2
2.3 EARTHQUAKE RESISTANT DESIGN
Seismic design is rather more complicated than conventional other-loading design
conditions in that it requires decision-making skills in following guidance formulae which
makes it an art as much as a science. Due to the several uncertainties inherent in earthquake
engineering, many decisions throughout the design process are based on engineering
judgment. Starting with the objectives, the designer has to decide on the objectives
depending on the expected use of the structure and according to the stakeholders
requirement about economic savings, and therefore the objectives can include added
performance criteria to the main life-safety objective. Then, according to the chosen
objectives and coordinating with architectural designs, the designer makes a judgment on the
appropriate amount of trading strength with ductility and chooses a lateral force-resisting
system to satisfy those requirements, whether a frame or shear wall or a combination of both.
Finally, depending on the regularity of the structure, the design budget and the expected
seismicity, the designer selects a means of representing the earthquake input, whether ground
motion time records or response spectrum, and in view of that applies a suitable analysis
approach.
2.3.1 Seismic design objective
The basic seismic design objective is satisfying the design equation that the computed
supply of the structure should be more than the corresponding demands imposed by the
earthquake, with a degree of functionality defined by social and economic factors, and with
an acceptable reliability in the realm of a very uncertain event. In this way seismic design
should achieve the requirements defined at its onset of collapse prevention and damage
limitation, by following a widely-accepted strategy. Also earthquake design should ensure
successful performance of buildings by ensuring the presence of a complete and fully inter-
connected lateral-force resisting system providing a competent load path, and controlling a
failure mode.
40
Chapter 2
2.3.1.1 Design criterion: supply greater than demand
Seismic demand is defined as the effect of the earthquake on the structure, while
supply is the ability of the structures to resist that effect without failure. Therefore, the
supply represents the response of the structure to the demand. Capacity, on the other hand, is
the maximum capability of a structure to resist an earthquake effect without failure. Due to
the dynamic nature of earthquake loading, the demand and supply of the structure are
strongly coupled, making the design equation not simple to satisfy. The supply of a certain
structure varies depending on the ground motion. And because inelastic behavior is inevitable
in a seismic event and is usually permitted on purpose for economic reasons, the supply and
demand are a mix of many control parameters not just strength and not just ductility.
Essentially an earthquake demands deformation of the building which can be resisted by
strength and stiffness or allowed in a ductile controlled way depending on the ductility of the
structure. Therefore the supply can be strength or ductility or a combination of both
depending on the level of damage accepted as defined by functionality and cost constraints.
Ductility may be difficult and expensive to achieve for some types of materials and structural
members and in this case the structure is designed to supply enough strength to resist ground
motion. Similarly for structures whose functionality dictates high performance during an
earthquake, such as nuclear facilities and hospitals, damage is avoided and therefore ductility
is not allowed and the supply is mainly strength. Nevertheless, in normal cases, there is
always a mix between ductility and strength on both the supply and demand side. The usual
trend in Egyptian construction practice is to provide supply more on the end of strength due
to the difficulty and the unfamiliarity with special ductile reinforcement detailing.
2.3.1.2 Design requirements
The main requirement of seismic design, as reflected in ECP-201 (2008), is avoidance of
collapse and life safety in case of a major earthquake, satisfying the ultimate limit state. A
second requirement of damage limitation in structural and nonstructural components, in case
of a minor frequent earthquake, to satisfy the serviceability limit state, is not directly involved
in the design process, while just being checked as an evaluation criterion after detailing of the
41
Chapter 2
structure, through checking drift requirements under a reduced response than that of the
design earthquake (limit state).
The role of design is to provide the correct combination of the control parameters namely
stiffness, strength and ductility to achieve the previous requirements, and to provide the
desired economy. Stiffness is the most relevant parameter to satisfy the requirements of
serviceability under the minor frequent earthquake. Also strength, controls the level of
inelastic excursion, and thus reflects on the damage limitation state. The amount of strength is
chosen depending on the limit allowed by the stakeholder according to the function of the
building and economic considerations. Therefore the amount of strength acts as a fuse to
restrict the seismic forces to this certain limit, after which the response is inelastic and the
ductility of the structure controls the behavior for satisfying the collapse prevention
requirement under the severe earthquake.
2.3.1.3 Design strategy
It follows from the previous discussion that the actual interaction of supply and demand
is that by limiting the supplied strength to a certain value, the demand of the earthquake for
strength is restricted to the strength supplied, and the rest of the demand becomes ductility
demand to be satisfied by a high ductility capacity in the structural members, thus achieving
better economy and better predictability of behavior. The philosophy of seismic design
follows this same interaction in a bottom-up approach. A lateral force resisting system is
chosen and its ductility supply is assumed. Then the seismic forces based on elastic (strength)
demand are reduced by this ductility supply to give reduced forces for which the members are
designed for strength, and this requires just linear analysis simplifying the design process.
This is first phase of member proportioning, and it involves capacity design concepts to
control modes of failure. The second phase of seismic design is checking serviceability
requirements and member detailing to provide the assumed ductility supply.
42
Chapter 2
2.3.2 Defining the design-basis earthquake
2.3.2.1 Accounting for uncertainty
The aforementioned design equation satisfying the exceedance of supply over demand of
the earthquake can only be satisfied with an acceptable degree of risk because of the
uncertainties inherent in both the supply and demand side. On the supply side, uncertainties
due to discrepancy between actual material properties and section dimensions and those
modeled in design, and variations due to human errors in design and construction quality are
usually accounted for in a statistical framework. On the demand side, however, accounting
for unpredictability is less reliable because there is a lack of a reliable-size and quality
database of earthquakes. Therefore the earthquake design value is usually derived using a
probabilistic approach, because this approach covers the likelihood of an actual earthquake
exceeding the design ground motion. Various entities involved in the construction of a
structure, for example, the designer, contractor and owner, usually have different degrees of
risk acceptance for the structure because this level reflects on the cost of construction.
The current procedure of the Egyptian code as well as many other seismic codes is to
incorporate this degree of risk acceptance as a probability of exceedance of the design
earthquake in the service lifetime of an ordinary structure approximately taken as 50 years,
The probability that a larger magnitude earthquake occurs during the lifetime of a structure is
smaller than the probability of occurrence of smaller earthquakes, therefore using recurrence
formulae, the number of earthquakes having a specified magnitude or greater can be
estimated, providing the annual probability of occurrence of this specific magnitude
earthquake. This earthquake magnitude is converted to some corresponding spectral value (as
will be explained in the next section), usually peak ground acceleration. Then assuming a
fixed predefined target annual probability usually 10% in 50 years, the design earthquake is
defined. This can also be reflected as the return period which is the period in years after
which there is a statistical probability of occurrence of an earthquake with the same
magnitude as the design earthquake. The return period is calculated as approximately equal to
the inverse of the annual probability of occurrence, for example a 10% in 50 years probability
would mean having a return period of 475 years. The higher the return period considered the
43
Chapter 2
less the risk considered in the design, but the less economical the structure is, because it
corresponds to a stronger seismic event.
2.3.2.2 Definition of seismicity and the design earthquake
The choice of magnitude of earthquake that represents the seismicity of the region and
thus the design earthquake is done by a seismic hazard analyst. For the case of code drafting,
probabilistic seismic hazard analysis (PSHA) is performed to represent zones of earthquake
occurrence. The seismicity is assumed to be uniformly distributed over these specified source
zones. The study should take into account magnitudes, geology, epicentral distances and
other parameters of the earthquake source producing the predominant contribution to risk as
well as past strong motion recordings typical of the predominant earthquake sources. All
possible earthquake scenarios (all possible magnitude, location and ground motion
probability level combinations) are considered, and the ground motion relation are computed
for each possible by regression analysis to determine the coefficients that determine the
equation of ground motion. The uncertainty as assumed (Section 2.3.2.1) is introduced in the
form of recurrence formula, and so each ground motion has a specified probability of being
exceeded in a given time period, as discussed before. When higher reliability in determining
earthquake inputs is required, a deterministic approach can be employed, where ground
motions are computed based on individual earthquake scenarios (each having a single value
magnitude, location, and ground motion probability level) for each seismic source and the
largest ground motion from any included scenario is considered the design ground motion.
This can be used when the potential for significant earthquakes in the site is not accounted for
in the regional seismicity model, and is related to site-dependent design spectrum as
discussed hereinafter. The results of a PSHA are seismic hazard curve relating the desired
level of probability (usually 10% in 50 years as discussed before) with a ground motion
parameter that can be:
Peak ground acceleration on stiff soil design values for each zone
Design spectral values (usually at 0.2 sec and 1 sec ) for each zone
A complete uniform hazard curve for each zone
44
Chapter 2
Each will be discussed in the following Section as relevant. National agencies provide these
values usually in the form of design hazard maps for each zone.
2
2.3.2.3 Code representation of earthquake action
For the sake of earthquake representation for design purposes, response spectra are
favored because they bring together under one representation the characteristics of ground
motion and the structures response which are the most important aspects for designers.
However, using a single past earthquake response spectrum to predict the expected ground
motion is inappropriate because prediction is basically recreation of past observed events and
the response spectrum for each past earthquake is unique and cannot be similar to another
earthquake even if they have the same maximum value of ground acceleration and measured
on the same site. Hence the response spectrum to be used in design must cover a range of
ground motions possible at the site. Also this design response spectrum has to be smooth to
avoid large variations in spectral acceleration with small variations in the period that can
result from a jagged plot of an earthquake response spectrum as was previously shown in
Figure 2.1. A Smooth design spectrum is basically created by one of the following ways: i.
processing of site specific ground motion time histories ii. statistical relationships iii.
empirical relationships and iv. code stipulations.
i. Site-specific Design Response Spectrum
The first design response spectrum was suggested by Biot (1941) who postulated that
when we possess a collection of earthquake spectrums at a given location, it is suggested
that a simplified envelope should be used as a standard spectrum for the purpose of design in
that region. This is only possible in sites where there is a large array of actual ground motion
2
In ECP-201 (2008) two return periods are considered: 475 years (10% in 50 years) and 95 years (10% in
10 years or 50% in 50 years). These are equivalent to the hazard levels of FEMA 273 of a rare event and a
frequent event. The design earthquake is based on a single event which is the 475 years return period
corresponding to the no collapse requirement, and correction factors
1
are given to correct the design base
shear for different types of building importance according to their intended use after the earthquake. Also for
checking inter-story drifts, correction factors are given to correct the resulting displacement from the 475
return period design earthquake to the 95 return period corresponding to the serviceability checks (ECP-201,
page 150).
45
Chapter 2
time histories recorded by strong motion instruments, and sites of similar seismicity that can
be represented by this group of earthquakes. The time histories are normalized to a fixed
ground motion parameter for example peak ground acceleration, to remove peaks and valleys,
and then their response spectra are constructed and compiled and then averaged by some
statistical method. Usually, the "mean" or means plus one standard deviation design
spectrum is developed and it has a smooth shape. The coefficient of variation can be used to
establish a spectrum with a desired probability level.
This method is the most accurate, however due to its dependence on the availability of a
large and reliable database of actual ground motions and its applicability only to the sites
where those ground motion were recorded and compiled, it is only used for especially
important structures and in limited sites which has a complicated soil condition that cannot be
covered by other types of design response spectrum (Taranath, 2005). When there is a lack of
historical ground motions and site-specific design spectrum is dictated by other
circumstances pertaining to the structure importance, ground motion records from sites of
similar seismicity are used, or artificial ground motion records are developed using numerical
methods that simulate the earthquake rupture process and the seismic wave propagation to the
site. Once constructed this site-specific response spectrum is used to design buildings on the
site itself or sites of similar seismicity. The definition of seismicity involved is usually the
result of a deterministic seismic hazard analysis, based on known earthquake sources.
ii. Statistically Derived Design Spectrum
With the increase in the size of earthquake database, many researchers developed
response spectral attenuation relationships that relate elastic design spectra at sites of
different soil conditions and tectonic environment, in the same way as ground motion
parameters are related by attenuation relationships. Ground motions records are categorized
based on compatibility of seismicity, and for each group, the design response spectrum is
developed by normalizing and averaging of all response spectra, then by statistical regression
analysis of the created design spectra for different classes of ground motion, attenuation
relationships are created relating each period spectral value to several parameters like
earthquake magnitude, soil conditions and epi-central distance. Then at any site, given the
design earthquake magnitude and distance, a whole design response spectrum can be
46
Chapter 2
constructed by inputting these parameters into the region-specific attenuation relationships at
each spectral value.
iii. Empirically Derived Idealized Design Spectrum
Due to the complexity in dealing with real ground motion spectra and the limited number
of records available a few decades ago, many researchers like Housner (1970b), Seed et al.
(1976), Newmark et al. (1973) and Newmark and Hall (1982) proposed empirical methods
for developing elastic design response spectrum from estimates of ground motion parameters
and based on the observation that all spectra have a distinctive shape. Of particular interest is
the idealized design spectrum created by Newmark and Hall (1982) which has been used,
with some modifications based on engineering judgment, extensively in structural design in
many seismic codes worldwide. Based on the observation that a response spectrum can be
roughly divided into three regions which are the short period region where spectral
acceleration is maximum and nearly constant, the intermediate period region where spectral
velocity is maximum and nearly constant, and the long period range region where spectral
displacement is maximum and nearly constant, they proposed a simple method for
constructing a design spectrum by generating a baseline curve of constant maximum ground
motion parameters in the three period-dependent regions described, and then generating the
design spectrum by amplifying the baseline curve by amplification factors which are based
on statistically derived spectra recorded on firm soil sites, and different amplification values
are provided for the median and median plus one standard deviation spectra. Newmark and
Hall's structural response amplification factors can also be used to change the damping value
of other spectra, such as those generated using attenuation relationships and code based ones.
These type of design spectra do not incorporate the important effects of earthquake
magnitude and characteristics on spectral shape, while they just include this effect indirectly
in the peak ground acceleration.
iv. Code-stipulated Design Spectrum
There are two kind of code-stipulated design spectrums, depending on the shape of the
spectrum and the consistency of the hazard assumed at all points of the spectrum: hazard
point-anchored design spectrums, and uniform hazard spectrum
47
Chapter 2
(a) Hazard point(s)- anchored design spectrums
Most codes, including the Egyptian code, assumes a shape for the design spectrum then
scales that spectrum to reflect the seismicity of the region (as defined by the design bases
earthquake), soil conditions and importance factors for different categories of structures
use.
Spectral shape:
The basic shape of design spectrum adopted by codes are usually based on a typical
Newmark and Hall's spectrum, except that there are only two spectral regions of
interest -- constant acceleration and constant velocity. The constant displacement
domain of the response spectrum is not included because commonly structures do not
have a long period that falls into this range. In order to account for different tectonic
features, for example, deep versus shallow earthquakes, different shapes are provided.
And soil coefficients are used to define the basic shape, which represent in a
simplified way the effect of different soil conditions on the shape of the spectrum as
first developed by Seed, Ugas and Lysmer (1976), and presented in Figure 2.11
Figure 2.11 Effect of soil conditions on spectral shapes (Seed, Ugas, and Lysmer, 1976, courtesy of
Seismological Society of America)
48
Chapter 2
Anchoring point(s):
The anchoring point of the adopted spectral shape is used to reflect the seismic hazard
in the region, and depends on the output of a PSHA. Either the zero-period horizontal
peak ground acceleration on rock sites is used as a single anchoring point, which is
the case for EC8 and ECP-201 (2008), or two anchoring points which are the spectral
ordinates at 0.2 and 1.0 seconds, as representative of short and long-period spectral
accelerations, as is the case of the 2003-NEHRP provisions and American codes. The
latter method have a greater appeal due to the higher reliability of the considered
uncertainty at more representative spectral points (e.g., Bommer and Pinho 2006,
Karakostas et al. 2007 ).
(b) Uniform hazard spectrums (UHS)
If for a given return period, the ground motion spectral acceleration is measured from
many hazard curves that relates the desired probability to a series of spectral periods, and
plotted at their respective spectral periods, the result is what is termed, a uniform hazard
spectrum. This is equivalent to anchoring the design spectrum at all points with the same
assumed probability; however in this case, the shape of the spectrum is not the same as a
typical Newmark-Hill spectrum. Figure 2.12 depicts an example of a uniform hazard
curve. This is relatively new method for characterizing seismic hazard for design. It has
the benefit of providing a consistent probability of exceeding the ground motion at any
period, however, since unifying the probability requires mixing ground motions from
different earthquakes to cover the whole range of spectral periods (short periods
controlled by nearby moderate magnitude earthquakes while long periods controlled by
distant large magnitude earthquakes), UHS does not characterize the spectrum of any
single earthquake, limits its application to only response spectrum analysis as opposed to
time history analysis because generating ground motion records to match the UHS would
be unrealistic and over-demanding. The typical Newmark-based code spectrum usually
provide multiple spectra for each important source, as is the case for type(1) and type(2)
design spectrum of the Egyptian code. Uniform hazard curves
49
Chapter 2
Figure 2.12 I llustration of a Uniform hazard spectrum
2.3.3 Analysis methods
Seismic codes allow different methods for analysis of structures according to their
importance and simplicity in layout and elevation. These methods differ mainly in
representation of the earthquake input, as well as the manipulation of the resulting responses.
Starting by the most accurate, seismic analysis methods are:
2.3.3.1 Linear and nonlinear time-history analysis
This type of analysis uses ground motion time-history acceleration records as the applied
loading which is the closest simulation of reality. The model used can be elastic (linear time-
history analysis) or inelastic (nonlinear time-history analysis). The rigor of the method lies in
employing the latter detailed nonlinear structure models that can produce results of relatively
low uncertainty. Responses are obtained by integration of the solution of the equation of
motion at each time step, while changing stiffness properties at each step in case of a a
nonlinear model. Despite the expected accuracy of results due to the as-real representation of
earthquakes, the response (displacement or force) is usually sensitive to the individual ground
motion used and therefore several analyses are usually required and the response is usually
calculated as (1) the maximum of responses due to three ground motions, or (2) the mean
value of responses due to seven ground motions.
50
Chapter 2
In order to reflect the code definition of the design-basis earthquake, the ground motion
records utilized in time-history analysis need to match the design response spectrum provided
by the code over the period range of significance for the structure. There are three general
approaches to developing spectrum-matching ground acceleration time-histories:
1. Selecting real past records whose spectrum matches the design spectrum: and which can
be simply scaled to improve the fit without changing its waveform and relative frequency
content.
2. Modifying real records to match the design spectrum: by adding wavelets, either in the
frequency domain or the time-domain.
3. Completely synthesizing artificial earthquakes: which although not similar in shape to real
records, provide best fit to design spectrum and therefore reduce the variability in results.
The expected variation in results based on the choice of ground motion, as well as the
liability to errors due to complexity and length of the analysis, and processing of a lot of input
and output data, make time-history analysis more suitable for research applications, and they
are usually prescribed by building codes only for buildings of unsymmetrical configuration or
of special importance. More about nonlinear time-history analysis will be presented as
relevant in Chapter 3.
2.3.3.2 Nonlinear static analysis
This is a simplified nonlinear analysis procedure in which a pattern of increasing lateral
forces is used to induce forces and deformations, and the total force versus displacement at all
incremental load points are plotted to define the capacity curve of the structure. This can be
then combined with a demand acceleration-displacement response spectrum to define
maximum response. Or alternatively an equivalent SDOF model with changing stiffness
properties along each incremental load can be used to embody nonlinear response.
51
Chapter 2
2.3.3.3 Response spectrum method
This is a linear dynamic procedure that permits taking multiple modes of response of a
structure into account. It is the main method of analysis in ECP-201 except for very simple or
very complex structures. The seismic input is modeled using the code response spectrum. In
order to include features of nonlinear behavior, the elastic response spectrum is modified by
dividing its ordinates by the R factor. The rationale behind this will be explained in the next
type of analysis. The response of structures from response spectrum modal analysis is
computed for each mode separately and then added together by some appropriate means like
square-root-sum-of-the-squares (SRSS) or complete-quadratic-combination (CQC)
2.3.3.4 Equivalent static load method (ESLM)
Despite being the most approximate method relative to the previously mentioned
methods, the equivalent static method is a central concept in seismic design and is still the
most widely used in all seismic codes and standards, due to its efficiency and simplicity. In
this method, the earthquake effects are represented by an equivalent static lateral load
distribution all-over the building height. The loads applied are defined by the design response
spectrum at the fundamental period of the building which the method assumes the building to
be predominantly vibrating with. For this to be true buildings need to be relatively low-rise
and with no irregularities to avoid rotational modes.
To account for nonlinear effects due to yielding, and making use of the equal
displacement rule as will be discussed in Section 2.4.1.1. that states that the displacement of
an inelastic and elastic system are equivalent for long-period SDOF systems, the code allows
reduction of the design spectrum and consequently the lateral forces by a force reduction
factor (FRF), referred to as R in ECP-201(2008) and many other codes, on the basis that the
extra cost by designing for a higher force, is not justified from a deformation demand point of
view. In other words, it is uneconomical to design to resist the complete load due to the
design earthquake, when a reduction in load would result in the same displacement and thus
damage of the system. As previously brought up in Chapter 1, this reduction results in
dragging the response of the structure into the inelastic region and therefore accounts for the
52
Chapter 2
energy dissipation that is preferred in the real structure. Inelastic action is also favored
because it ensures a predictable mode of failure by ductile action. The R-factor chosen
depends on the assumed ductility of the system which is in its turn related to the expected
maximum displacement. In that case, the displacement determined from the elastic analysis
need to be amplified by the R-factor to go back to the elastic structure displacement, and then
multiplied by another factor to account for the difference in displacement due to inelasticity.
These two factors combined make the DAF, which is the main interest of this research. The
second factor in particular which relates the inelastic displacement to its elastic counterpart is
of prime importance. This is equivalent to the ratio of DAF to R-factor as shown on the
idealized force-deformation diagram in Figure 2.13.
Figure 2.13 Illustration of force reduction and displacement amplification factors
The Egyptian code specifies a value of 0.7 for the ratio between DAF and FRF, where it
provides the following equation (8-25 in the code) for estimation of maximum inelastic
displacement:
d
s
= 0.7 R d
e
(2.9)
53
Chapter 2
where:
d
s
is the displacement of a point of the structural system induced by the design seismic
action
d
e
is the displacement of the same point of the structural system, as determined by a
linear analysis based on the design response spectrum
R is the force reduction factor
This ratio is consistent with the provisions of UBC (1997) code in the United States, and is
the only provision in ECP-201 that doesnt follow the EC8 which set it as 1. And still with
this reduction and despite some studies (e.g Freeman (2000)) indicating this 0.7 value has a
technical basis; its over-conservatism is commonly doubted in the Egyptian design
community. The following Section describes various research work that studied this ratio in
various forms.
2.4 LITERATURE REVIEW
The current study divides the review of available literature on estimation of inelastic
displacement demands under seismic loading into two main broad categories. The first
category is conducting a general review of the available literature and methods of estimation
of inelastic deformations, and the second category is dedicated to a detailed analysis of
previous work and findings about lateral drift estimation and evaluation of the displacement
amplification factor in a code-based context, which is more relevant to the work at hand.
2.4.1 Literature review related to estimation of maximum inelastic
displacement demand from maximum displacement demands of
elastic systems.
Due to the availability of numerous relationships for inelastic displacement to elastic
displacement ratios in the literature, this part of literature aims to describe the different types
54
Chapter 2
and models of major research efforts in this field and the reader is referred to the mentioned
studies for additional information on specific findings.
2.4.1.1 For SDOF systems
Because time history analyses are usually not practical in a design context, , seismic
displacement demands are usually specified by idealizing structures as SDOF and the
maximum inelastic displacement demands are approximated from the maximum response of
linear elastic single-degree-of-freedom (SDOF) using design linear elastic response spectra as
discussed before. Therefore there is a particular inclination in the research realm to estimating
maximum inelastic displacement demands from the maximum displacement demand of linear
elastic SDOF systems. Miranda, a very active researcher in the area of estimating maximum
inelastic displacement demands, categorized the methods of estimating inelastic displacement
demand of SDOF as (Miranda, 2001):
i. direct method in which maximum inelastic displacement demands are directly related to
maximum displacement of the same linear elastic system having the same lateral stiffness
and same damping coefficient though a factor defined as the inelastic displacement ratio.
ii. indirect method or equivalent linearization method in which the maximum
displacement is estimated as equal to the maximum displacement of an equivalent linear
elastic system with lower lateral stiffness (higher period of vibration) and with higher
damping coefficient than those of the system for which the maximum inelastic
displacement is being estimated.
There are several studies that evaluated and compared the different methods and
underlying developed models, for example among others the research by Jennings (1968),
Iwan and Gates (1979), Iwan (1980), Hadjian (1982), Miranda and Ruiz-Garcia (2002a,
2002b) and Akkar and Miranda (2005). However, because the current study focuses on the
general relationships between maximum inelastic displacement and the underling important
factors, the details of applicability and quantitative assessment of the accuracy of each
relationship is beyond the scope of this study. The literature review herein uses the same
categorization by Miranda (2000), starting with the earlier of them.
55
Chapter 2
i. Indirect method (Equivalent linearization method)
Jacobsen(1930) was the first to introduce the concept of equivalent viscous damping by
obtaining approximate solutions of the steady forced vibration of damped SDOF systems
with linear forcedisplacement relationships and damping forces proportional to the nth
power of the velocity of motion when subjected to sinusoidal forces, without change in the
stiffness than the real system. The equivalent viscous damping ratio employed was based on
equating the energy dissipation per cycle of the original damping force to that of the
equivalent damping force. As an extension to the concept of equivalent viscous damping, the
same author later (Jacobsen, 1960) applied it to yielding SDOF systems by considering some
selected period shifts each having different values of equivalent viscous damping .
For selecting the period shift in an equivalent linear system, many studies exist that
relate the stiffness of the equivalent linear system to that of the original one. The method
initially proposed by Rosenblueth and Herrera (1964) is the most implemented one especially
in displacement-based design (non-linear static procedures), which is based on the secant
stiffness at maximum deformation. This method is also referred to as the geometric stiffness
method. In this study, the researchers estimated the equivalent viscous damping ratio on
equating the energy dissipated per cycle of steady response to harmonic excitation in the non-
linear and equivalent linear SDOF systems and concluded that the circular frequency of
vibration of the equivalent linear system is given by
=
2
=
0
+
2
1
1
=
0
+ 0.21
1
= 1 +0.121( 1)
0.939
and
=
0
+ 0.0587( 1)
0.371
.
In a more recent study, Kowalsky (1994) extended on the secant stiffness method for
defining the period shift, by employing Takeda hystertic model,a nd employing nonlinear
time history analysis under earthquake loading, to derive an equation for the equivalent
viscous damping ratio. The equivalent damping ratio was given by
=
0
+
1
1
1
with modifications for including post-yield stiffness.
ii. Direct methods (displacement coefficient method)
Using this method, the maximum displacement response of the inelastic SDOF system
max, is estimated as a product of the maximum displacement of a linear elastic system e,
with the same lateral stiffness and same damping coefficient as that of the inelastic system
(i.e. k0 and 0) times a displacement modification factor, C, as follows:
..(2.10)
57
Chapter 2
This is the method employed by force-based seismic codes to evaluate the displacement
demands of structures relative to that resulting from a linear analysis, which represents the
core of the current research. Many studies and relations exist that relates the maximum
inelastic displacement to that of the corresponding linear system, as referred to as the
inelastic displacement ratio.
The first and most influential research work in the area of relating inelastic displacement
demands of SDOF systems to their elastic counterparts is the work by Veletsos and Newmark
(1960). In their work they used three earthquake records to numerically analyze assumed
elastoplastic models of SDOF and concluded that the maximum inelastic displacement is
equal to the maximum elastic displacement, a rule referred to as the equal displacement
rule and forms the basis of most seismic design codes. They also proved that this rule
doesnt apply in the short-period range of structures that fall below the period separating the
constant-acceleration and constant-velocity part of the elastic response (referred to as the
characteristic period) where the inelastic displacement considerably exceeds its
corresponding elastic value. This study among other studies (e.g. Veletsos et al, 1964)
presented the foundation for the well-known Newmark and Hall (1972) method for
estimating inelastic response spectra from elastic response spectra. In this method the
inelastic displacement ratio varies depending on the spectral region in which the initial period
of vibration of the SDOF system is located. The main conclusions of Nemwark and Hall
(1972) are that:
For long-period structures, regardless of the ductility demand, the total displacement of the
elastic and inelastic systems are the same, and C= 1. This is referred to as the equal
displacement rule.
For intermediate- period structures, the velocity of vibration of the elastic system and the
inelastic system is the same and so is the total energy absorbed , and =
21
. This is
referred to as the equal energy rule.
For short-period structures, the acceleration of vibration of the elastic system and the
inelastic system is the same, and so their strength demand is the same, and C=..
Figure 2.14 explains the equal displacement rule and equal energy rule as would be
depicted on a force-deformation idealization of a structure. Figure 2.15 shows the results of
58
Chapter 2
the study in the form of the different displacement modification factors for each ductility
level. It should be noted that the inelastic displacement ratio for short-period and
intermediate-period structures is higher than one according to Newmark and Hall (1982)
Figure 2.14 Illustration of Newmark and Hall (1982) (a) equal displacement rule and (b) equal
energy rule for long and intermediate period structures, as applied on a supply curve
Figure 2.15 Graphical depictions of the Newmark and Hall displacement modification factors
across a range of periods on a demand curve.
59
Chapter 2
After the research outcome of Newmark and Hall (1982), many studies were conducted
to investigate the relationship between the displacement modification factor and the period of
vibration and ductility in the system. These relationships are usually referred to as R--T
relationships, although a more straightforward relationship would have been obtained using
the displacement modification factor.
In 1984, Shimazaki and Sozen built on the Newmark and Hall (1982) study by including
more hysteretic models, either bilinear or of Clough type (Clough and Johnson, 1966), and
conducted numerical analysis using El-Centro earthquake record. Their results confirmed
Veletsos and Newmark previous findings, and enhance it by relating the amount of increase
of inelastic displacement over elastic one in the short period range to the type of hystertic
model and to the lateral strength. Their conclusions were confirmed by Qi and Moehle
(1991). In 1998, Whittaker, Constantinou, and Tsopelas, performed a similar numerical study
but used 20 earthquake records on a model created by Bouc-Wen (Wen, 1979) and built up
mean and mean plus one standard deviation plots of maximum inelastic to elastic
displacement ratios for different strength values.
The recent works by Miranda and his co-researchers (Miranda, 1991, 1993a, 1993b)
have provided a large range of inelastic displacement ratios for use in the direct method. In
his first efforts in 1991 and 1993, Miranda analyzed 30,000 SDOF systems of elasto-plastic
material model using 124 earthquake ground motions recorded on different soil types, and
developed different ratios of maximum inelastic to elastic displacement for three types of soil
conditions, as well as evaluated the period value where the equal displacement rule applies.
His continuing research in 2000 (Miranda, 2000), was extended to provide different inelastic
displacement ratios for different earthquake magnitudes, epi-central distances as well as soil
conditions. In that study, it was concluded that inelastic displacement ratio was not
significantly affected by earthquake magnitude or distance to source. Also the study
presented a simplified unified single expression that relates the displacement modification
factor to period and ductility, unlike the several expressions used byNewmark and Hall
(1982). However the trend of both methods was the same.
Many studies investigated the effect of many other factors other than ductility and period
of vibration. The effect of strength softening was investigated by Miranda and Akkar (2003).
60
Chapter 2
Inelastic displacement plots for structures on soft soil were developed by Ruiz-Garcia and
Miranda (2004, 2006).
Krawinklers studies are also abundant in this field. Nassar and Krawinkler (1991),
Rahnama and Krawinjkler (1993) and Seneviratna and Krawinkler (1997) carried out studies
on SDOF systems similar to those of Miranada, and additionally included the effect of
strength degradation or stiffness degradation, and investigated the use of bilinear, Clough, or
of pinching type. A major finding is the single unified expression developed by Krawinkler
and Nassar (1992) that relates the inelastic displacement ratio of SDOF systems to the
fundmantal period and force reduction factor as C =
1
1 +
1+
+
1
1 where Ts is a
soil-dependant coefficient given as 1.05 for NEHRP site class D, and a,b and c are site-
dependent constant coefficients, and the relationship developed by Chopra and
Chintanapakdee (2004) which is = 1 +
1
1
+
1
where
=
1
+1 and Tc is the period at the start of the acceleration sensitive region of the
response spectrum given as 0.41s for NEHRP soil type D, and a, b, c and d are constant
coefficients. The latter group pf researchers proved that the equal-displacement rule
overestimates inelastic displacements in the vicinity of Tc , when Tc is predominant in the
vibration like in soft soils. Chenouda and Ayoub (2008) evaluated and calibrated these three
expressions for estimating inelastic displacement ratios by including the effect of strength and
stiffness degradation.
Employing degradation models in estimation of inelastic displacement ratios was also
present in many other studies. In their research, Gupta and Kunnath (1998) employed 15
ground motion records to conduct a numerical analysis study on SDOF systems using three-
parameter degradation model. While the degrading model, used by Song and Pincheira
61
Chapter 2
(2000) in another study on inelastic displacement ratios, was dependant on the number of
cycles only without consideration of hysteretic energy dissipated or any collapse potential.
Chenouda and Ayoub (2008) developed new inelastic displacement ratios for range of SDOF
systems of varying periods, applying 4 bins of ground motion records each comprising 20
earthquakes and utilizing a modified Clough model (Clough and Johnson, 1966) for concrete
together with a bilinear model for steel. They relied on the hysteretic energy dissipation for
presenting degradation with an eight-parameter criterion based on the work of Rahnama and
Krawinkler (1993) that considered strength degradation, unloading stiffness degradation,
accelerated stiffness degradation and cap degradation.
2.4.1.2 For MDOF systems
Fewer studies have been presented for estimation of inelastic displacement demand of
MDOF systems. This is primarily because for the sake of design usually structures are
idealized as SDOF systems and therefore the methods of SDOF previously mentioned are
applied. For example, several recently proposed displacement-based methods (Moehle
(1992); Kowalsky et al., (1995); Calvi and Kingsley (1995); Priestley (1996), Miranda
(1999), Chopra and Geol (1999) and Fajfar (2000)) use the response of linear elastic SDOF
systems to estimate the maximum inelastic displacements in bridge and building structures..
Following the same categorization used for SDOF, the methods available in the literature are
those in which displacement of a linear system is multiplied by factors or displacement is
calculated for an equivalent linear system.
The method of equivalent linear idealization is more popular in case of MDOF systems
because it takes into account the effect of higher modes in representation of stiffness of the
equivalent system. Alonso, Miranda and Santa-Ana (1996), Miranda (1997), Seneviratna and
Krawinkler (1997), Gupta and Krawinkler (2002), and Medina and Krawinkler (2005), used a
multi-stage method in which an equivalent SDOF system is first established and its inelastic
displacement demands are estimated same as for SDOF systems, then multiplicative factors
are estimated based on nonlinear time history analysis or pushover analysis to bring close the
behavior of the SDOF to the MDOF system. This is considered as an extension to the secant
stiffness method used for SDOF (Rosenblueth and Herrera, 1964), and for MDOF, the
62
Chapter 2
method is referred to as the substitute structure method. The stiffness properties of the actual
structure and the substitute structure are related as (EI)i, substitute = (EI)I,actual /i, where i
depends on the acceptable damage of a particular element i.
Of particular importance in the study of MDOF systems displacements is the work by
Gupta and Krawinkler (2000) which provides the basis of the displacement coefficient
method used in American codes. In this method, the researchers propose and evaluate a
method in which the maximum story drift demands are approximated through a series of
modificatuob factors applied to the first mode spectral displacement. These factors include
(1) MDOF modification factor the relates the elastic spectral displacement at the first mode
period of the structure to the elastic roof drift demand of the MDOF structure, neglecting P-
delta effects; (2) Inelasticity modification factor that relates the elastic roof drift demand to
the inelastic roof drift demand, neglecting P-delta effects (similar to the displacement
modification factor for SDOF systems); (3) P-delta modification factor that accounts for
geometrical nonlinear effects on the inelastic roof drift; (4) storey drift modification factor
that relates each storey drift demands to the roof drift demand; and (5) element deformation
modification function that relates the storey drift demand to the elements plastic deformation
demands.
Moreover, studies on MDOF studies involve investigating factors that were not included
in SDOF systems for example the effect of irregularity and torsional effects (Fajfar et al.,
(2005); Marusic and Fajfar (2005); Kosmopoulos et al. (2003); Mola and Negro (2005);
Molina et al. (2005), Panagiotakos and Fardis (2006) and Kosmopoulos and Fardis (2006)).
However, these studies involved the more accurate methods of static pushover, nonlinear
time-history analyses, or full-scale testing of some MDOF systems, which are beyond the
scope of this study, and therefore they didnt impinge on the approximate methods for
estimation of maximum inelastic displacement, or the relation between inelastic and elastic
displacement.
63
Chapter 2
2.4.2 Literature review related to estimation of maximum inelastic
displacement in a code-based context (DAF as related to the R-factor)
There are many studies questioning the values of the displacement amplification factor
employed in various codes by claiming the over-conservatism of the resulting inter-storey
drift values relative to computed values using more accurate methods of analysis and shake-
table tests, as well as the excessive separation distances as proved by pounding studies. And
although calibration studies of force reduction factor used in seismic codes are abundant,
there is very limited research suggesting tuned values for the displacement amplification
factors to be used in the framework of the equivalent static method widely used by seismic
codes.
Uang (1991) used the relationship between force reduction and ductility developed by
Newmark and Hall (1982) to formulate an explicit expression for the displacement
amplification factor, referred to in his study as Cd consistent with NEHRP , which is the ratio
between maximum inelastic displacement max, and elastic displacement at the design level
force (reduced by R), s. Recognizing the relationship between maximum displacement and
displacement at yield which is defined as ductility, =
y
; and defining the structural
over-strength factor as the ratio between the actual yield strength of the system and the design
(assumed yield) strength =
which is equivalent to
, the displacement
amplification factor was defined as
1
: is importance factor for the building, equals 1.2 for office building
R: is the force reduction factor prescribed a value of 5 for moment resisting frames of
limited ductility.
83
Chapter 3
Figure 3.4 Design horizontal response spectrums and governing equations for the study zones
For calculation of base shear, the code allows computing the fundamental period from a
modal analysis, (using a computer program) provided that this value does not exceed an
upper bound calculated as 1.2 multiplied by the period approximated by the practical formula
(equation B-1) provided in the code (ECP-201,2008), given by:
T
1
= C
t
H
3/4
(3.3)
Where
T
1
is the fundamental period of the structure in seconds
C
t
: is a factor depending on structural system &material taken as 0.075 for concrete
frame buildings
H: is total height of the building measured above foundation level, in metres
A comparison between the fundamental periods computed and those approximated by
the practical formula as well as the upper bound specified by the code is presented in Table
3.4. As shown, the calculated base shear values were always bound by the restrictions of the
code. This is because upper bound equations are deliberately formulated and skewed to result
84
Chapter 3
in shorter and more conservative fundamental periods of structures, which will result in
higher design base shears. Also the computer model doesnt take in the stiffening effect of the
nonstructural infill walls and cladding as well as structural members not part of the lateral
force resisting system, which are all present in real buildings (Jacobs, W., 2008). In the same
line of specifying a lower bound on the calculated base shear, the code provides a lower limit
on the ordinate of the design response spectrum as 0.2g, as shown previously in Figure 3.4.
Table 3.4 Comparison between computed and code approximated building fundamental period
Frame reference
Fundamental period
from 3-D model on
SAP2000
Fundamental period
from code approximate
equation B-2
Upper bound on
calculated period
F01 0.47 0.171 0.205
F04 0.822 0.484 0.580
F07 1.24 0.736 0.883
F10 1.63 0.961 1.154
3.3.1.3.2 Vertical distribution of base shear
The calculated base shear forces were distributed vertically on each prototype frame
according to the distribution of weights (corresponding to the masses used in modal analysis)
along the height pursuant to the code provided equation:
i i
i b
i i
1
WZ
F = .F
WZ
n
i =
(
(
(
(
(
.(3.4)
Where;
F
i
: is the horizontal seismic force for story i.
W
i
: is the weight of story i.
Z
i
: is the story height at story i.
These horizontal forces were applied at each floor at the center of rigidity of the floor,
with an additional moment to account for accidental torsion (due to unpredictable live load
and variations of material properties) by assuming displacement of the center of mass each
85
Chapter 3
way from its actual location by a distance equal to 5% of the dimension of the building in
plan perpendicular to the direction of applied horizontal forces.
3.3.1.3.3 Design load combinations considered
The prototype buildings were designed to resist all possible load cases that combine
gravity and seismic loads. According to ECP-203 (2007), they were designed for the higher
of:
U
1
= 1.6D + 1.4L.....,,,,..(3.5)
U
2
= 1.12D + oL + S.(3.6)
Where DL: dead load
L L : live load
UL : ultimate load
S : Seismic load
o : live load factor taken as 0.5 for public and office buildings.
3.3.1.3.4 Design and Detailing
Moment resisting frames of limited ductility are not prescribed much different detailing
requirements than non-seismic structures, according to ECP-203 volume 6 (2007). Table C.1
in Appendix C summarizes the design and detailing requirements provided by the Egyptian
code for limited-ductility MRFs, (ECP-203, 2007, volume 6). The design and detailing
features of the columns of the representative buildings in the three zones are elaborately
illustrated in Figures 3.5 to 3.8. Also the beams detailing is presented for the seven storey
buildings as a sample in Figure 3.9, which shows the elevation together with the cross
sectional dimensions and details. The design and detailing considerations of beams are
extended for use in the design for the other buildings as shown in Figures C.1 to C.4.
Regarding the design of elements, the following should be noted:
Bottom reinforcement was extended till the support, in order to account for the moment
reversal at the end of beams due to earthquakes.
86
Chapter 3
The only capacity design rule applied is that resulting from the prescribed reduction in
effective flexural stiffness of members where the reduction for beams is higher than that
for columns.
The four prototype buildings were designed by fixing their cross-sectional dimensions
(which is the proportioning results of the design in Zone 3)
4
, and adding steel
commensurate to the added moment due to lateral loads. This helps to provide common
comparison basis among the different zones without the effect of the difference in
fundamental period, if the cross sections were changed. In the parametric study, some
cases were studied for the effect of re-proportioning members versus adding more steel.
The change in amount of reinforcing steel in columns in a building among the different
zones was done, as much as possible, by changing the reinforcing bar diameters rather
than changing the number of reinforcing bars, in order to maintain almost the same
transverse reinforcement configuration, presented by the number of stirrups and their
layout. This was done also to provide a fair comparison in terms of the confining effect of
concrete as will be discussed in Section 3.3.4.2.2.
4
Middle-seismicity zone lateral loads were chosen for proportioning the elements, not just the gravity load, in
order to avoid over-reinforcing the elements, if too small cross sections are chosen relative to the straining
actions or consistent minimum reinforcement in all cases if too large cross sections are used.
87
Chapter 3
(scaled for illustrative purposes)
Figure 3.5 Elevation layout (top) and column cross-section detailing (bottom) for the one-storey
frame
88
Chapter 3
(scaled for illustrative purposes)
Figure 3.6 Elevation layout (top) and columns cross-section detailing (bottom) for the four-storey
frame
89
Chapter 3
(scaled for illustrative purposes)
Figure 3.7 Elevation layout (top) and columns cross-section detailing (bottom) for the seven-storey
frame
90
Chapter 3
(scaled for illustrative purposes)
Figure 3.8 Elevation (top) and columns cross-section detailing (bottom) for the ten-storey frame
91
Chapter 3
92
Chapter 3
3.3.2 STAGE II: Displacement Analysis
This stage represents the core of the current research, where it demonstrates and
compares the displacement results a designer would obtain from a building already
strength-designed according to the code prescriptions for the equivalent static force method
using the code provided estimation based on the results from the same method of equivalent
static force, and using nonlinear time-history analysis method considered to be the most
accurate method of analysis. Since the latter method focus on large displacement analysis in
the nonlinear range, a quantitative assessment of this effect requires analytical models
capable of capturing the nonlinearity of the structure under dynamic reversed cyclic loading.
The varieties of available inelastic models are presented herein.
3.3.2.1 Overview of inelastic modeling techniques
According to the level of discretization considered in the idealization of the modeling of
the structure and the idealization of inelastic behavior, analytical models can be divided into
three categories defined in decreasing discretization as follows:
3.3.2.1.1 Microscopic models
Analyses using these models are also referred to as Continuum Analyses. Stress-strain
behavior is monitored throughout each and every point of the structure. These are by far the
most accurate and powerful models, however their associated excessive computational effort
and difficulty in learning usually restrict their use for modeling individual members or sub-
assemblages or at the maximum very simple structural configurations (El Tawil and Deierlin,
1996). Examples of the common finite element programs based on the continuum approach
are: ABAQUS, ADINA, ANSYS, DIANA, NASTRAN. Most of these programs do not have
a standard section database, and do not have a convenient means of applying ground motion
records and load combinations. For these reasons, these categories of finite element programs
were not considered for use in the current study.
93
Chapter 3
3.3.2.1.2 Macroscopic models
These models are usually referred to as the distributed plasticity models, or the fiber
analysis method. In this method, each member may be modeled using several elements
depending on the degree of refinement required, and the element behavior is monitored at
different cross-sections which are in turn discretized into a relatively large number of small
fibers. For reinforced concrete structures, sections are usually discretized into steel, confined
and unconfined concrete fibers, with different material constitutive models to describe the
fibers stress-strain relationship. The accuracy of the model increases as the number of points
at which nonlinear behavior is monitored increase.
The fiber method makes use of the assumption that plane sections remain plane after
bending. Therefore using the relevant input constitutive model, fiber stresses are calculated
from the fiber strains which are monitored during the entire multi-step analysis taking into
account the migration of the position of the section neutral axis during the loading history.
Moment-curvature relationships of the sections are in their turn obtained by integrating the
stress-strain response of the fibers over the cross-sectional area. This is followed by
integrating the moment-curvature relationship of the sections along the length to obtain the
moment-rotation response.
This type of model automatically accounts for concrete cracking and growth in crack
length, however cracking is considered to be smeared and normal to the member axis as a
result of the plane section assumption. Local buckling of steel bars can be modeled by
providing a steel constitutive model that degrades the structural properties of the steel
elements which reach a critical buckling stress. Also fiber models account for interaction
between axial force and bending moment. Examples of programs that employ the fiber
approach are: OpenSees, ZEUS-NL, SeismoStruct and DRAIN-2D and 3-D.
3.3.2.1.3 Phenomenological models
These models are also referred to as lumped-plasticity (plastic hinge) models. Inelastic
force deformation behavior is monitored and assumed to be restricted to specified locations at
94
Chapter 3
ends of elements (plastic hinge location) and modeled as a zero-length hinge with hysteretic
properties, as shown in Figure 3.10(b). It does not account for reduction in stiffness of
members and for the interaction between axial force and bending moment; however it has the
benefit of using a single element to model one structural member and being computationally
efficient due to the concise form of the stiffness matrix. Example of the programs widely
used for this kind of modeling are: SAP2000, ETABS, Perform, and DRAIN-2D and DRAIN
3-D.
Figure 3.10 (a) Modeling of a plastic hinge using fiber method (macroscopic model)
5
; (b) modeling
of a plastic hinge using lumped plasticity method (phenomenological model) (FEMA, 451)
As a compromise between the accuracy of micro-models and the computational
efficiency of phenomenological models, it was chosen to use the macro-models based on the
fiber approach in the present study. This choice was reinforced by the reliability of this
modeling method in predicting response similar to experimental and full-scale tests as
reported by many researchers (e.g. Broderick, 1994; Pinho, 2000).
5
The fiber method can model the whole structural member in the same discretized way as shown, not just the
plastic hinge region.
95
Chapter 3
3.3.2.2 Analysis program
3.3.2.2.1 Reason for choice
The selection of an analysis program that uses the fiber approach involved choosing
between (OpenSees, ZEUS-NL and SeismoStruct)
6
, based on the following criteria with their
relative importance presented as a percentage:
1. Reliability based on validation in literature and developer reputation (40%)
2. Relevance of capabilities to the current study of global displacement analysis (40%)
3. Ease of use and quality of interface (20%)
3.3.2.2.2 History and capabilities
ZEUS-NL is a finite element structural analysis program developed by the Mid-America
Earthquake Center, at the Newmark Laboratories of the University of Illinois at Urbana-
Champaign, as part of its core research program of Consequence-Based Engineering
(Elnashai et al., 2003). It is based on the analysis packages ADAPTIC (Izzudin and ElNashai,
1989) and INDYAS (Elnashai et al., 2000) developed at Imperial College in London. It
provides a completely visual and efficient means for nonlinear analysis of two- and three-
dimensional steel, reinforced concrete and composite structures under static and dynamic
loading, taking into account the effects of both geometric nonlinearities and material
inelasticity. The program is capable of representing the spread of inelasticity within the
member cross-section and along the member length through utilizing the fiber approach
(Elnashai et al. 2002). It also has the feature of predicting large displacements of frame
members by imposing equilibrium in the deformed state of the structure and hence it can
represent geometrical nonlinearity and P-delta effects. Each element has its own local
coordinate system for defining force-deformation properties and for interpreting output.
Analysis options are: constant static loading, conventional pushover, adaptive pushover,
eigen-vlaue, linear and nonlinear time-history and incremental dynamic analysis using the
augmented ZBeer post-processing utility. For dynamic analysis, the program can take either
displacements or accelerations varying independently in the time domain and applied at
6
DRAIN-2D and 3-D were excluded from the selection because their reliability in applying fiber modeling is
often doubted by researchers while they usually rank-high in applying the lumped plasticity approach (FEMA
451, 2003)
96
Chapter 3
supports or at nodes. Available output includes deformation across the element, displacement
at nodes, internal forces at joints of elements and support forces. Input for the program is
described as relevant in Sections 3.3.2.4 and 3.3.2.5
Table 3.5 Comparison between considered fiber modeling soft-wares
Program OpenSees ZEUS-NL SeismoStruct
Criteria
S
c
o
r
e
Explanation
S
c
o
r
e
Explanation
S
c
o
r
e
Explanation
Reliability
(40%)
3
Widely used & verified
in the literature
3
Widely used &
verified in the
literature
1
Limited use in the
literature
Developed by Pacific
Earthquake Engineering
Research (PEER)
Center at University of
Berkeley, LA
Developed by Mid-
America Earthquake
(MAE) Center at
University of Illinois,
UC
Developed by a Private
company (Seismosoft)
Relevance
of
capabilities
(40%)
3
Geometrical
&material nonlinearity
3
Geometrical &material
nonlinearity
3
Geometrical &material
nonlinearity
Nonlinearity in cross-
section and along
beam length
Nonlinearity in cross-
section and along
beam length
Nonlinearity in cross-
section and along beam
length
Ease of
use
And
graphical
interface
(20%)
1
Based on C++
2
Visualization of input
model and nodes
3
Visualization of input
model and nodes as well as
results of input data (cross
sections, detailing and
constitutive models)
Plot , tabulated and
video options for
results
Plot, tabulated and video
options for results
No graphical interface Post-processing of
results (ZBeear utility)
Monitor performance
criteria (color-coded form)
Score (/3)
2.6 2.8 2.2
Special
Strengths
Flexibility (Open source)/can
be combined with other
programs
Ability to pause and re-start
analysis
Exceptional ability to model
soil structure interaction,
range of models for soil
behavior
Exceptional graphical interface
Special
Weaknesses
No slab elements and
distributed loads
No slab elements and
distributed loads
Displacement-
based*element
* Regarding element formulation, fiber models can be divided into force-based & displacement-based
elements. The latter are less accurate than the first, because the beam displacements are expressed as functions
of the nodal displacement using a prescribed shape function (usually cubic), which is not accurate because of the
assumption of cubic displacements meaning linear curvatures. This issue is solved by using several elements in
a single structural member, thus increasing the number of global degrees of freedom. While in a force-based
frame element, the beam section forces are expressed as a function of the nodal forces through force shape
function. This is exact within the small-deformation Euler-Bernoulli beam theory, and therefore one force-based
element per structural member can be used.
97
Chapter 3
3.3.2.2.3 Behind the black-box
In both static and dynamic analyses, permanent loads are applied at the start and iteration
to equilibrium is performed. Other types of loads (proportional lateral loads or ground
displacement or acceleration) are applied incrementally afterwards. In eigenvalue analysis,
Lanczos algorithm is employed for determination of natural frequencies and modes of
vibration. In the case of time-history analysis, the non-linear equations of motion are solved
iteratively and integrated either using the unconditionally stable Newmark time integration
method or the Hilber-Hughes-Taylor algorithm (Broderick et al., 1994). This may re-quire
re-forming and re-solving the stiffness and damping matrices in each time step. Comparisons
between the two schemes confirmed the similarity between their results at the design
earthquake, as confirmed from the study of Broderick et al. (1994), therefore the more
common Newmark integration algorithm was employed in the present study. The iterations
are carried out until the solution converges. Convergence is specified according to either
force-moment or displacement-rotation criteria, depending on whether the analysis
increments from one point to the next converged point of a force-displacement relationship,
using an increment of force or of displacement. Additionally, automatic time-step reduction is
performed if convergence is not realized during analysis. Moreover, ZEUS-NL utilizes an
updated Lagrangian form for the large displacement formulation, where deformed member
axes are used to derive member deformations. A detailed description of available material
models and elements in ZEUS-NL is beyond the scope of this study. Only the element
formulations and material models used in the present work will be briefly presented when
applicable.
3.3.2.2.4 Validation
ZEUS-NL finite element code in its present or previous forms of ADPATIC and
INDYAS have been validated and comprehensively tested by a large number of researchers
over the past 20 years. Examples of verification at Imperial College of London at the member
and structure levels include among many others the work by Izzudin (1991), Madas and
Elnashai, (1992), Elnashai and Elghazouli (1993), Broderick and Elnashai (1994), Martinez-
Rueda (1997), Lee(1999) and Pinho (2000. The further improvement and verification of the
program was extended at the University of Illinois at Urbana-Champaign, USA, for example,
98
Chapter 3
the validation against full scale test results done by Jeong and Elnashai (2005), and against
SAP2000 (Elnashai et al., 2004). This research work helps to prove the stability and
robustness of the program so that it is deemed appropriate for the current research.
For added confidence, the author performed a simple verification exercise by comparing
the Eigenvalue modal analysis results for a 2-D model of the four-storey building under
study, using ZEUS-NL, SAP2000 and OpenSees. The results are displayed in Table 3.6,
where the three programs provided very close results with a small difference due to applying
different element formulations and computing procedures. The mode shapes from the
programs were exactly the same which ensures the reliability of the analytical tool.
Table 3.6 Verification of modal analysis results (natural period in seconds) of ZEUS-NL against
SAP2000 and OpenSees
ZEUS-NL SAP2000 OpenSees
Mode 1
0.777 0.783 0.761
Mode 2
0.248 0.266 0.229
Mode 3
0.141 0.136 0.146
3.3.2.2.5 Units
ZEUS-NL limits the user to using the metric units of N-mm-sec. Because in the USA,
concrete compressive strength is measured by cylinder tests rather than cube tests as of
common practice in Egypt, the assumed f
cu
had to be converted to its cylindrical value, where
the cylinder strength was taken to be 20% less than the cube strength as described by ECP-
203 (2007). This results in cylindrical compressive strength (f
c
) of 20 N/mm
2
. This same
value was used in the calculation of confinement factors according to Section 3.3.2.5.
3.3.2.3 General procedure and assumptions
Taking advantage of the symmetry of the buildings, a 2-D model of one frame was
chosen for displacement analyses, in order to simplify the post-processing of results when
99
Chapter 3
time-history analysis is employed. The first interior frame was selected in this case because it
is the most critical one when drift-response is under question. The reasons for that are that
they are more affected by P-delta effects due to larger gravity loads, and that their design
indicated larger cross sections to satisfy the higher gravity loading which resulted in having
higher stiffness and thus higher share of the total lateral load on the space frame.
The output results of stage I of design for gravity and seismic load combinations were
used for two purposes:
- Horizontal distribution of the base shear forces among the different frames (ratio of
base shear of one frame to base shear of the whole structure) to be used for
equivalent lateral force analysis.
7
- Cross-section dimensions and detailing of the various members, based on strength
design to be used for both equivalent lateral force and non-linear time history
analyses.
Three types of analyses for determining displacement demands were performed:
1. Equivalent static force (preceded by modal eigen-value analysis)
2. Linear time-history
3. Nonlinear time-history
All analyses were performed using ZEUS-NL in order to have a common platform for a fair
comparison of results, and using the same assumptions presented in stage I (Section 3.3.1),
unless mentioned otherwise. Gravity loads were included in all analyses.
It should be noted that stage I ended at arriving at straining actions and designing
sections accordingly, while in stage II, the 2-D models cross sections were updated to reflect
final design, and the displacement analysis is performed on this updated model using both
methods of equivalent lateral force and time-history analysis. Therefore results from
7
Based on the rigid diaphragm assumption, the shear in any story is to be distributed among the vertical
components in proportion to their contributions to the lateral stiffness of the story This contribution of any
component is the product of its lateral stiffness and the square of its distance to the center of resistance of the
story. Using the program SAP2000 to determine the apportionment of base shear on the studied 2-D frame
results in more accurate (and critical) results than hand calculations because it includes the effect of the slab (T-
sections versus L-sections) on the rigidity of the frame and consequently its share of total lateral load.
100
Chapter 3
equivalent lateral force method on the 3-D structure (stage I) are not comparable to those
from stage II where stiffness properties and accordingly fundamental periods and equivalent
lateral loads are not the same.
Another difference in results of equivalent static force method between stage I and stage
II stems from the new assumption in the current stage of ignoring accidental torsion. This is
again important in order to have a fair comparison between the two methods, where it is
practically unfeasible to include accidental torsion in time-history analysis where several
different dynamic analyses would have to be conducted with masses located differently that
would result in different dynamic properties and base shears for each run, necessitating undue
effort in post-processing of results to find the maximum response. Thats why according to
most US seismic codes
8
, it is recommended to treat accidental torsion as a pure static torsion
load in a separate load condition whose results can be appropriately combined with other
static and dynamic loads (Wilson, 1996). Pertinent to this recommendation, the effects of
accidental torsion would cancel out when comparing results of the two methods, and thus
were ignored altogether.
Because ZEUS-NL doesnt have an element capable of applying distributed load, gravity
load distributed pattern was simulated by applying several loading points on a beam. Three
nodes were defined dividing each beam element into quarters. Equivalent point loads were
calculated using the concentrated load equivalents factors in the Table 5-16 of the third
edition of LRFD of AISC (2001), as shown in Figure 3.11 (Bai and Heuste, 2007). This
discretization also helps increase the accuracy of the results. In order to reduce the size of the
mass matrix in dynamic analysis, the number of lumped masses was reduced by placing them
at beam-column connections instead of the several loading points along the beams.
For determining equivalent static base shear forces, the upper bound on the computed
fundamental period of the structure was ignored, together with the lower bound on site-
related spectral acceleration, as described in Section 3.3.1.3.1. This also serves to control the
variables in the comparison between the static and dynamic analyses results, so that both
would be based on the same computer-generated periods and the same basis of seismic action
8
The Egyptian code followed in this study doesnt provide any provisions regarding handling accidental torsion
in case of dynamic analysis.
101
Chapter 3
which is the code design response spectrum unaltered, and thus the difference in results
would be based only on inelastic behavior. The alternative option of imposing an upper limit
Figure 3.11 Equivalent point loads applied on beam members
on the fundamental period would entail corresponding modeling of infill walls in the
nonlinear dynamic analysis. This option was not employed because infill walls usually get
damaged at low drift values and therefore their contribution to stiffness ceases. Also it should
be noted that infill walls alter the response of the frames in an unpredictable way that should
be evaluated on a case-by-case basis depending on whether they are full height or partial and
their actual distribution. Moreover although infill walls create more stresses on frame, this
stress is created by interfering with the frame movement and restraining drift thus acting as
compression struts that impart loads to the frame. Therefore the effect of infill walls on the
displacement behavior of the building is not as critical as the stress behavior, and although
the upper bound equations on the fundamental period are a must for strength design, they are
just an option for drift calculations (Jacobs, w., 2008), as per the recommendations of ASCE
7-05 and IBC (2000). The results in Chapter 4 show that this method results in much less
over-conservatism in the code estimated value of displacement than the currently employed
method of restricting the computed fundamental period to a certain value.
The general procedure and output of this stage are shown in Figure 3.12. As mentioned
before, there were five main different assumptions in stage II displacement analysis:
1. Two-dimensional modeling of the prototype buildings.
2. Accidental torsional effects were ignored.
3. Lumped masses at beam-column intersection.
4. Code lower-bound limits on base shear (period and spectral acceleration) were ignored.
5. P-delta effects included.
102
Chapter 3
Figure 3.12 Illustration of general procedure for displacement analysis and assessment critertia
103
Chapter 3
3.3.2.4 Modal Analysis and Equivalent Static Method
This step followed the same procedure described in Section 3.3.1.2 and Section 3.3.1.3
and applying the assumptions in Section 3.3.1 and 3.3.2.3. For the sake of brevity, only
element formulations and cross-section properties are described herein, while the overall
modeling followed almost the same description provided later in non-linear time history
analysis. Cubic elastic elements with a cubic shape function for estimating transverse
displacement were employed as shown in Figure 3.17. This formulation accounts for
geometric nonlinearities, but doesnt account for material inelasticity. Member stiffness were
based on the cracked sections as specified by ECP-201 (2008) as 0.7 I
g
for columns and 0.5 I
g
for beams, this was accomplished by reducing the moduli of elasticity for each member with
their respective factor. Slabs contribution to beam stiffness and strength was reflected by the
effective flange width of the T-section, taken as 1150mm according to stage I design. The
cross-sections and materials employed are shown in Figure 3.13, and using dimensions as
determined from Stage I of design.
J c(a)mgjhgcjhsgdjhsgdhjdshjjjgd d (b)
Figure 3.13 Elastic model cross sections and materials for (a)beams ;(b)columns
104
Chapter 3
The fundamental period, base shear ratio and calculated total base shear for each of the
four prototype buildings in the three study zones are presented in Table 3.7
Table 3.7 Calculation of base shear (a) using computed T
1
(b) using code upper limit on T
1
Ref. T
1
(sec)
Reactive
weight(KN)
Zone 1 Zone 3 Zone 5B
C
d
F
b
(KN) C
d
F
b
(KN) C
d
F
b
(KN)
F01 0.321 1026.5 0.053 54.86 0.080 82.29 0.160 164.59
F04 0.777 7299.8 0.029 211.44 0.043 317.16 0.087 634.32
F07 1.192 13277.9 0.019 250.54 0.028 375.81 0.057 751.63
F10 1.629 18192.1 0.010 185.14 0.015 277.71 0.031 555.41
(a)
Ref. T
1
(sec)
Reactive
weight(KN)
Zone 1 Zone 3 Zone 5B
C
d
F
b
(KN) C
d
F
b
(KN) C
d
F
b
(KN)
F01 0.205 1026.5 0.090 92.39 0.135 138.58 0.270 277.16
F04 0.580 7299.8 0.039 283.05 0.058 424.58 0.116 849.15
F07 0.883 13277.9 0.025 338.38 0.038 507.57 0.076 1015.14
F10 1.154 18192.1 0.024 436.61 0.036 654.92 0.072 1309.83
(b)
3.3.2.5 Dynamic Time History Analysis Method
In dynamic analysis, the fundamental equation of motion is used to determine the
response of the structural models, and is given by:
+ + =
.(3.7)
where:
[M] = Mass matrix
{a} = Acceleration vector
[C] = Viscous damping matrix
{v} = Velocity vector
[K] = Structural stiffness matrix
{u} = Displacement vector relative to the ground
a
g
= Ground acceleration
This is done in an iterative process until equilibrium is achieved and the results are
integrated through a step-wise numerical integration scheme. The time-step of analysis is
105
Chapter 3
specified as the same time-step of the applied loading, and if convergence is not achieved the
analysis time step is automatically reduced to a smaller step. The user define the time-step for
applied loading, and this was decided by starting with an initial load-step equals to the ground
motion record time-step, then repeating the analysis using reduced load-steps until the output
results are no more affected. In case of nonlinear analysis, the stiffness matrix is revised at
each solution time-step.
The problem with time-history analysis is that the results are only applicable to the single
earthquake record employed in the analysis. Using a different earthquake record would
produce completely different results. Therefore several time-history runs have to be
employed to have a more general picture of the response. In the current case studies, seven
ground motion records were used for each building model in the three seismic zones, and the
results were averaged, pertinent to the provisions of ECP-201 (2008). The code also allows
utilizing three ground motion records and taking the maximum of their results, however many
studies point to the bias created in the results by taking the maximum response related to a
single earthquake action.
Since linear time history analysis uses the same material and cross section models as in
modal analysis and equivalent static loading analysis, only the nonlinear model is described
hereafter.
3.3.2.5.1 Assumptions
The general assumptions from Section 3.3.2.3 and as applicable from Section 3.3.1 are
used, in addition to the following:
1. There is no history of nonlinear deformation where at initial conditions (before nonlinear
static analysis), displacement and velocities are zero.
2. Damping ratio is assumed 2% in the first mode and 5% in the second mode.
3. Beam-column connection is modeled without rigid links or shear joints.
This is validated by the results of a study by Joeng and Elnashai (2004) which
compared the results of a numerical moment-resisting frame model on ZEUS-NL
106
Chapter 3
employing: Model #1: NO rigid links and NO joint shear; Model#2: rigid links and NO
joint shear; Model #3 NO rigid links and joint shear; Model#4: with both rigid links and
joint shear, to those from a full-scale experimental model, and proved the viability of the
models with no rigid links since they are closer to experimental results, as shown in
Figure 3.14. This was attributed to better accounting for plastic hinge locations due to
bond slip.
4. Shear deformation of members are ignored.
This is validated by the results of a comparison of experimental studies with
numerical models on the older form of ADAPTIC, which proved that the effect of
inclusion of shear modeling is minor for members controlled by flexure as those
employed in the present study, that they can be ignored (Elnashai et al., 1999; Lee,
1999). These results are shown in Figure 3.15.
Figure 3.14 Validation of assumption of neglecting beam-column rigid connection & shear at joint (show
top displacement results at Center of mass, 0.15g PGA test) ( After J oeng &Elnashai, 2004)
107
Chapter 3
Figure 3.15 Validation of assumption of neglecting shear deformation (shows effect of inclusion of
shear deformation modeling on cyclic response) (After Lee, 1999)
3.3.2.5.2 Numerical Model
i. Building model
The buildings were modeled as a two-dimensional model on ZEUS-NL. Structural nodes,
represented by blue squares, were defined to connect element, represented by black lines as
shown in Figure 3.16. Each structural node was restrained out-of-plane and the ground
support nodes were fully fixed except in the x-direction of loading to allow accelerations at
the support. In addition to the meshing previously described for allowing the gravity load to be
included as equivalent point loads, each member was divided to reflect changes in reinforcement
detailing in the section as per Figures 3.5 to 3.9 and C.1 to C.4, and to ensure proper
discretization for accurate capturing of inelastic action, by providing shorter elements near
108
Chapter 3
member edges where dissipative zones are expected. These meshing criteria resulted in using
10 and 7 elements to model each beam and column respectively, as shown in Figure 3.16.
Figure 3.16 Meshing of the seven-storey building elements
ii. Elements models
Three different types of elements were used in the model:
- Beam-column elasto-plastic element, to model frame elements
- Rayleigh damping element, to model viscous damping of the structure
- Lumped mass element , to model masses at beam-column joints
1. Beam-column elasto-plastic element
A 3-D cubic elasto-plastic element formulation was utilized to model the spatial
behavior of column and beam elements (Izzudin and Elnashai, 1990). This element type
employs a cubic shape function to calculate the transverse displacement as shown in
Figure 3.17. The cubic element stiffness matrix is integrated using second order Gaussian
quadrature, which shows the importance of employing short lengths elements size near
beam-column connections where maximum inelastic behavior is expected.
109
Chapter 3
Figure 3.17 3-D cubic formulation for elastic and elasto-plastic beam column element
Two integration Gauss points per element are used for the numerical integration of the
governing equations of the cubic formulation as shown in Figure 3.18. At those two Gauss
points, each cross-section was divided into a large number of monitoring areas chosen as
200 in order to ensure accurate modeling of the distribution of material nonlinearity
across the section. The cross sections were described using RC rectangular sections (rccs)
and RC T-sections (rcts) for columns and beams (and slabs) respectively. The
discretization of each cross-section is also shown in Figure 3.18 and Figure 3.19. It
accounts for the actual arrangement of longitudinal reinforcement as shown in the
detailing figures in stage I and the concrete cover taken as 25mm, while the arrangement
of transverse reinforcement is approximated through the constitutive material model as
will be explained later. The sectional stress-strain state of beam-column elements is
obtained through the integration of the material stress-strain at the fiber level, where a
bilinear elasto-plastic material model with kinematic strain hardening was utilized to
simulate the inelastic cyclic response of the reinforcement and uniaxial constant
confinement concrete material models to describe concrete behavior. Details of these
material models are presented later in this Chapter.
110
Chapter 3
Figure 3.18 Discretization of cross-sections at the two integration gauss points of each element
Figure 3.19 Cross sections used in modeling beams and columns
2. Rayleigh damping element
Rayleigh damping elements were chosen to model viscous damping in the structure
due to friction in opened micro-cracks of concrete and nonstructural elements
interaction, as described in Chapter 2, while hysteretic damping due to inelastic behavior
111
Chapter 3
and yielding is implicitly included within the nonlinear material model specified for the
frame elements that allows energy to be dissipated during cyclic loading. As previously
discussed, the non-hysteretic damping part is usually very small compared to energy
dissipation by inelastic action, that some researchers even strongly discourage its
modeling altogether (Wilson, 2001). Other researchers advise its employment in very
small quantities mainly for providing numerical stability where the viscous damping
matrix results in stabilizing the system of equations, and through the use of only stiffness
proportional damping, as discussed by Pegon (1996), Wilson (2001), Abbasi et al. (2004)
and Hall (2006), because mass-proportional damping usually generates excessive
unrealistic energy dissipation if a structure is insensitive to rigid body motion. Also in
the present study, since the fixation of the support of the building is released in the
direction of loading of earthquake, providing mass proportional damping will be relative
to the absolute velocity rather than the relative velocity and will result in erroneous
results.
For these reasons it was chosen to model equivalent viscous damping by stiffness-
proportional Rayleigh coefficient, assuming 2% of critical damping in the first mode and
5% of critical damping in the second mode, covering the modes of significant mass
participation.
9
It should be noted that the ECP-201 (2008) specifies using design
spectrum based on 5% critical damping in the first mode, however this percentage also
includes indirectly the effect of inelastic behavior, so 2% was considered similar to the
code provisions for wind load in which structures are expected to behave elastically.
Stiffness-proportional Rayleigh damping parameter was calculated based on the
periods of the structure in the two considered modes (Chopra 2000). The percentage of
critical damping ,, for a specific mode n, depends on mass-proportional damping
coefficient o
and stiffness-proportional damping coefficient |, as well as on the
frequency e
n
, as expressed by equation (3.8)
,
=
1
2
..(3.8)
9
Periods used in calculating the stiffness-proportional damping coefficient are calculated based on un-cracked
sections, because the stiffness matrix used in calculation of damping is updated at every time step, therefore
cracking is considered through the nonlinear analysis.
112
Chapter 3
By specifying the percentage of critical damping for two modes, i and j, the coefficients
can be computed by solving two equations in two unknowns. The stiffness proportional
damping is calculated following equation (3.9), and Table3.8 illustrates the values used
in the prototype buildings
=
2
2
.(3.9)
Table 3.8 Stiffness-proportional damping coefficients used in the four prototype buildings
Ref. T
1
*
(sec)
T
2
*
(sec)
e
1
(rad/sec)
e
2
(rad/sec)
1
2
|
F01 0.248 0.052 18.058 120.414 0.02 0.05 0.00080
F04 0.602 0.196 10.429 32.130 0.02 0.05 0.00303
F07 0.917 0.349 6.854 18.016 0.02 0.05 0.00550
F10 1.235 0.465 5.087 13.525 0.02 0.05 0.00732
*Un-cracked elastic member stiffness used in determining periods of structures
3. Lumped mass element
Masses were modeled as lumped masses at intersection of beams and columns
because the approximation of lumped masses has been proven to be quite accurate for
seismic analysis of symmetrical frames, with an added advantage of reducing
computational demand.
iii. Material models
1. Concrete model
The uniaxial nonlinear concrete model used in this study, shown in Figure 3.20, was
implemented by Madas and Elnashai (1992) adopting the constitutive relationship
formulated by Mander et al. (1988) which was proven by the authors to provide a good
balance between simplicity and accuracy. Several researchers have validated this model
and recommended its use over other available confinement models based on experimental
values (Kappos et al. (1998); Rossetto (2002)). Moreover, the cyclic rules were
significantly improved by Martinez-Rueda and Elnashai [1997] to allow the prediction of
113
Chapter 3
continuing cyclic degradation of strength and stiffness, as well as providing better
numerical stability under large displacements analysis.
Figure 3.20 Uni-axial constant-confinement concrete material model used in ZEUS-NL
The model employed is applicable to both confined and unconfined concrete, and for
various shapes of cross section and it accounts for increase in strength and ductility due to
confinement, taking into account the maximum confining pressure that can occur at yield
of transverse reinforcement. It assumes constant confinement pressure throughout the
entire stress-strain range, introduced on the model through a confinement factor that
scales up the whole stress-strain relationship.
The difference in behavior between unconfined and confined concrete as described by
Mander (1989) is depicted in Figure 3.21. Increasing confinement by providing closer
spaced transverse reinforcement increases the strain capacity of members, by restraining
the lateral expansion of the concrete, delaying buckling of longitudinal reinforcement and
enabling higher compression stresses to be sustained. (Mwafy, 2001). The confinement
factor is defined as the ratio of confined concrete strength (f
cc
) to unconfined concrete
strength (f
c
).
In the current study, each cross section is defined by two separate concrete material
models for the core concrete and cover concrete. The calibrating parameters and their
associated values used to fully describe this material model are tabulated as follows
114
Chapter 3
(Table 3.9). The calculation of confinement factor according to the arrangement of
longitudinal and lateral reinforcement for core concrete is described hereinafter.
Table 3.9 I nput parameters for concrete uniaxial constant confinement model
Parameter Description Values used
f
c
Unconfined compressive strength 20 N/mm
2
f
t
Tensile strength 2.2 N/mm
2
c
co
Strain at peak stress 0.002 (mm/mm)
K Confinement factor 1 (unconfined)
Table D.1 (confined)
Figure 3.21 Unconfined and confined concrete monotonic stress-strain behavior (after Mander et. al, 1989)
STEP 1: The effective confining stress (f
1
) that can be developed at yield of the transverse
reinforcement is calculated by:
f
lx
=k
e
.
x
.F
yh
.(3.10)
f
ly
= k
e
.
y
.F
yh
(3.11)
In x and y directions respectively, where:
x
and
y
: are effective section area ratios of transverse reinforcement to core
concrete cut by planes perpendicular to the x and y directions, respectively.
115
Chapter 3
K
e
: is a confinement effectiveness coefficient relating the minimum area of
effectively confined core to the nominal core area bounded by the centerline of the
peripheral hoops. This factor depends on the distribution of longitudinal steel and the
resulting tie configuration and spacing. A typical value is provided by pauley and
priestley (1992) as 0.75 for rectangular sections, which is used in this study.
F
yh
: is yield stress of the stirrups.
STEP 2: When the concrete core is confined by equal lateral confining stresses (i.e. f
l1
=
f
l2
) as is the case for the symmetrically reinforced square columns employed in this
study, equations (3.10) and (3.11) make one equation that defines f
l
, and the confined
compressive strength of concrete is calculated as:
=
1.254 +2.254 1 +
7.94
(3.12)
Therefore confinement factor is given by:
= 1.254 +2.254 1 +
7.94
(3.13)
The confinement factor calculated for columns of all prototype frames in zone 3 are
shown in Table D.1 in Appendix D. The confinement factor ranged between 1.19 and
1.44. The same confinement factors were used in all zones by maintaining the same
arrangement of stirrups in order to have a common basis for comparison. Beams were
assumed unconfined in all cases due to the limited effect of transverse reinforcement of
beams in improving flexural deformation capacity where they are mainly employed to
protect the beams in shear.
2. Steel model
A bilinear (elasto-plastic) model with kinematic strain-hardening was utilized to simulate
the inelastic response of steel longitudinal bars of the cross-sections of the RC beam-column
elements. employing this simple bilinear uniaxial relationship were proved to correlate well
with experimental results as reported by Bursi and Ballerini 1996,Salari et al, 1998, among
others. In this model, loading in the elastic range and unloading phase follows a linear
116
Chapter 3
function defined by a constant value which is the Youngs modulus of steel. In the post-
elastic range, a kinematic hardening rule for the yield surface defined by a linear
relationship to the initial stiffness is assumed as shown in Figure 3.22 (Elnashai and
Elghazouli, 1993; Elnashai and Izzudin, 1993). The values and parameters used in the
present work are shown in Table 3.10.
Figure 3.22 Uniaxial elasto-plastic steel model with kinematic strain-hardening used in ZEUS-NL
Table 3.10 I nput parameters for the uniaxial bilinear steel model with kinematic strain hardening
Parameter Description Values used
E Youngs modulus 205900 N/mm
2
o
y
Yield strength 360 N/mm
2
Strain hardening parameter 0.005
3.3.2.5.3 Representation of Earthquake Action
The selection of ground motion records to be used in time-history analysis represents the
most important and difficult stage in time history analysis, because of the sensitivity of the
structural response to any change in input earthquake characteristics combined with the great
uncertainty inherent in reflecting the actual seismicity of the region under study. The main
important characteristics of ground motions and their effect on structures were discussed in
117
Chapter 3
Chapter 2. Therefore in order to have meaningful results from inelastic time-history analysis,
the earthquake input has to be carefully selected.
Seismic design codes, including ECP-201 (2008), generally define ground shaking in the
form of a response spectrum that envelopes the average of 5% damped response spectra of
different earthquake scenarios. And in case of time-history analysis, it permits the use of:
a. Natural accelerograms recorded during earthquakes
b. Artificial records
c. Synthetic ground motions obtained from seismological models
Under the following conditions:
1. They match or exceed the design spectrum within a period range of interest
2. They reflect the seismological features of the area under study
3. A suite (group) of ground motion records are used to represent the different
earthquake sources in order to overcome some of the variability inherent: generally
for each suite 3 records considering their maximum response, or 7 records considering
their average response.
Because the current study aims to assess and compare the viability of the code estimate
of displacement demand for different buildings in three different zones designed according to
the code requirements, and because the design response spectrum provided in ECP-201 is to a
big extent based on that of EC-8 without any adjustment to reflect the seismological features
of Egypt, and its not within the scope of the present work to doubt the applicability of this
spectrum to Egypt, it was chosen to employ artificial records in conducting the time-history
analyses used in the displacement comparative analysis, with the single criterion of:
compatibility of the 5% damped elastic spectra of these ground motions with the code
spectrum used in the seismic design of the buildings. This again goes in line with the
requirement of unifying all variables used in the comparison by providing an equivalent
representation of earthquake action. Seven records, rather than three, were chosen for
analyses and the corresponding response averaged, as recommended by many researchers to
avoid bias in the results.
This choice of artificial records satisfied the first and third conditions specified by the
code as previously mentioned. In order to meet the second condition to some extent, zone 3
118
Chapter 3
building models were chosen for further refinement by applying real earthquakes that
represent the seismicity of the region. This also helps to overcome the problems associated
with using artificial ground motion records as will be described in the next Section.
i. Selection of artificial ground motion records
Artificial accelerograms were generated using the code SIMQKE (Vanmarcke and
Gasparini, 1979). This software constructs a time history record matching a user-defined
spectrum by building a power spectral density function from the smoothed response spectrum
and then producing sinusoidal signals of random phase angles and amplitudes. The program
then iteratively filters a series of white noise with a trapezoidal function of amplitudes in time
domain.
Seven ground motions were generated, all of duration 20sec, taking into consideration
variability of frequency and energy content as described by the maximum acceleration to
maximum velocity (A/V) ratio, as well as the earthquake predominant period, T
p
(The
various frequency content indicators were mentioned in Chapter 2, Section 2.2.1.2 (i)). This
variability helps reduce the bias in the displacement response. The software
SEISMOSIGNAL (Seismosoft, 2008) was used to evaluate the aforementioned parameters
for the generated records. Table 3.11 demonstrates the chosen artificial earthquakes and their
characteristics.
Table 3.11 Characteristics of selected artificial ground motion records
Earthquake reference Predominant period (T
p
) A
max
/V
max
ART-EQ1 0.26 11.23596
ART-EQ2 0.12 8.62069
ART-EQ3 0.2 13.88889
ART-EQ5 0.28 9.433962
ART-EQ7 0.18 11.76471
ART-EQ8 0.16 14.08451
ART-EQ9 0.22 10.52632
119
Chapter 3
The same set of records was used in zones 1, 3 and 5B, by scaling the records by the
corresponding PGA. This crude method of scaling is employed due to the fact that the
artificial earthquakes spectrums do not relate to real earthquakes, and are already matching
the shape of the design spectrum at all periods therefore there is no need to provide separate
scaling factors for each different height building depending on its fundamental period, also
serving to avoid changing factors in the comparison between buildings. Figure E.1 in
Appendix E shows the generated ground motion time-history records for zone 3, scaled by
0.15 and their corresponding 5% damped elastic spectrum as compared to the code Type (1)
response spectrum calculated for Zone 3 (PGA =0.15g) and soil type C. In terms of spectral
response, the selected records provide a good match and so are deemed in sound agreement
with the hazard level used in design. Figure 3.23 also shows the matching of the average of
the seven records 5% damped spectra with the code spectrum for Zone 3.
Figure 3.23 5% damped spectra for generated artificial ground motion records, and their
average compared to code design spectrum
120
Chapter 3
For ZEUS-NL input, the records were scaled by 9810 to be consistent with the software
units and by the importance factor (
1
=1.2) pertinent to the recommendations of ECP-201
(2008) following EC8 (2003). It should be noted that US codes (UBC, IBC, ASCE) follow a
different procedure, whereby the records spectra are matched to the design response spectrum
that include the importance factor in it. Although these two procedures are equivalent in the
linear range, they produce different response of structures when considering nonlinear
behavior, as will be proved later.
ii. Selection of real strong-motion records
Some researchers expressed doubts about the use of artificial accelerograms because
they have inadequate low frequency content and an excessive energy content spread all over
the length of the signal which is not typical of actual recordings of strong earthquake motion
and therefore can distort analysis response results (Werner, 1991). This can be readily
observed by comparing the artificial and real time-history ground motion records presented in
Figures E.1 and E.2 respectively in Appendix E. Real earthquakes normally have their
energy content relatively concentrated with peaks and troughs. This difference can question
the adequacy of results of nonlinear time-history analysis employing artificial earthquakes.
Also, reference to the discussion in Section 2.2.1.3, where the effect of site conditions on the
characteristics of ground motion and consequently on the response of structures was
discussed, the result of including possible magnitudes, epicentral distance and soil conditions
are worth investigating. Therefore the buildings in Zone 3 were chosen for refinement of their
earthquake input, due to the importance of this zone by having the highest seismic risk.
In order to obtain ground motion records that reflect the seismicity of Zone 3 as
represented by Cairo and other surrounding areas, the history of earthquake damage in this
area had to be studied. This indicated that the main hazard in Zone 3 is from nearby, shallow,
crustal earthquakes of small-to-moderate magnitude; this kind of earthquakes have caused
excessive damage in the past in other parts of the world (e.g. Bommer et al., 2001). Examples
of those events that affected Cairo are the M
s
5.8 event
10
of 1847 (Ambraseys et al., 1994)
and the M
s
5.4 Dahshour earthquake of October 1992 (Khater 1993; Badawy and Mourad,
10
Richter Magnitude Scale
121
Chapter 3
1994). Both earthquakes originated from the seismically active area to the southwest of Cairo
which is part of the Pelusiac (Eastern Mediterranean Cairo Fayoum) Trend (Neev et al.
(1982) and Kebeasy (1990). This area is located around 25km from the center of Cairo.
Although larger earthquakes are expected to occur from other seismological sources, their
distance is far enough that the resulting ground motions in zone 3 are unlikely to be of
importance.
The problem of using real earthquakes based on just the seismological features of the
area is that in most cases they will not match the design response spectrum, and thus they are
not suitable for the purpose of design. The predominant practice then is to select records
from available databases like PEER (PEER GM, 2007) or the European Strong-motion
database ESD (EC, 5
th
framework program) based on magnitude, distance, site conditions,
and other parameters that control the ground motion characteristics, then the selected records
are manipulated in order to match the code spectrum, using either frequency-domain methods
where the frequency content is manipulated (Gasparini and Vanmarcke 1976, Silva and Lee
1987, Bolt and Gregor 1993, Carballo and Cornell 2000) or time-domain methods where only
the amplitude of the recorded ground motions is manipulated. Scaling and matching can be
done either on a record-by-record basis using different scaling factors, or using their average
spectra scaled by one factor to match the design spectrum.
However, there is prevalent concern in the engineering community pertaining to the
practice of scaling records (Mehanny and Deierlein, 2000). For example, Han and Wen
(1994), Bazzuro , Cornell, Shoma and Carballo (1998), and Luco and Bazzurro (2005) all
questioned the scaling procedures that do not account for changes in ground motion
characteristics including Magnitude and frequency content, because they result in unrealistic
ground motions incompatible with physical observations. Even the widely accepted
procedure of structure-based scaling to match the spectrum at the fundamental period of the
structure, as suggested by Shome et al. (1999), introduces a bias since it reduces variability in
the resulting response as reported by Luco and Bazzurro (2005), and can be excessively
demanding. Moreover the spectrum matching becomes no more valid once the period of the
structure increases due to yielding (period elongation), as well as at other periods which can
be experienced by the structure due to higher mode effects.
122
Chapter 3
For the above reasons, un-scaled records were used in the present study, representative of
the seismicity of Zone 3 cities and matching the code spectrum almost constantly (using a
lower bound and upper bound) along a range of periods representative of the fundamental
periods of the structure. This also helps in making a fair comparison between the four
prototype buildings and provides closer simulation to reality by showing the effect of the
same un-scaled earthquake on different-height buildings without any bias. For this end,
because it is almost unfeasible to satisfy all criteria of code-spectrum matching and similarity
of seismological features by manual search (Beyer and Bommer 2007; Iervolino et al. 2008,
2009), a software package REXELv 2.61 beta (lervolino et al., 2009) was employed in the
current work.
11
This program allows searching for sets of seven records compatible in their
average with a pre-defined spectrum, and according to user-specified criteria reflecting
source seismological features that include Magnitude, epicentral distance and soil conditions.
The matching criteria applied are those prescribed in EC8 (2003) which are the same as ECP-
201 (2008). These criteria are, as outlined in the Egyptian code in Section 8.4.3.2:
a. A minimum of 3 accelerograms should be used;
b. The mean of the zero period spectral response acceleration values (calculated from the
individual time histories) should not be smaller than the value of a
g
S for the site in
question (S is the soil factor);
c. No value of the mean 5% damping elastic spectrum, calculated from all time histories,
should be more than 10% less than the corresponding value of the 5% damping elastic
response spectrum. (with the following addition only present in EC8: In the range of
periods between 0.2 T
1
and 2 T
1
, where T
1
is the fundamental period of the structure in the
direction where the accelerograms will be applied, however, this added restriction will
not make a difference in results because it gives a range that covers period elongation,
and higher mode effects)
The European strong motion database ESD (EC, 5
th
framework program) was used for
the search because it was used in formulating the EC8 spectrum, so it provides higher hope
for finding a match, and because it includes past earthquakes in Egypt. The seismological
11
A similar software called, The Design Ground Motion Library (DGML) was developed under a project
funded jointly by the California Geological Survey-Strong Motion Instrumentation Program (CGS-SMIP) and
the Pacific Earthquake Engineering Research Center-Lifelines Program (PEER-LL), at the University of
Berkeley, California, however it is not yet available for use, as of April 2010.
123
Chapter 3
criteria set were those corresponding to the previously mentioned seismicity, where
accelerograms need to have magnitude between 5 and 6 and epi-central distances less than
70km. Also because soil affects the ground motion by amplifying accelerations, and the
structural model did not include a soil model, the searched recorded were restricted to
stations located on soft soil conditions (Type C), as assumed in the structural analysis. The
set of earthquake records selected according to REXEL search and their related
characteristics as evaluated by SEISMOSIGNAL(Seismosoft) are shown in Table 3.12. It
should be noted that the PGA was not included as part of the search criteria, where it is
agreed among researchers that this criteria does not have a profound effect on structural
response as shown by Anderson and Naeim (1984), Anderson and Bertero (1987), Uang and
Bertero (1988), Bertero et al. (1991) and Naeim and Anderson (1993) . The comparison of
the average elastic spectra (5% damping) of the ground motion records with the code
spectrum (Site Class C) is shown in Figure 3.24, which shows that a very close spectral
match was achieved.
Table 3.12 Characteristics of selected real ground motion records
Name Region Year Station M
s
D
s
(km)*
Soil
Type
PHA
(m/s
2
)**
T
p
(sec)
d
(sec)***
Chenoua Algeria 1989 Cherchell 5.9 29 C 2.83 26.3 0.12 11.98
NE of
Banja
Luka
Bosnia
&herzegovina 1981
Banja Luka-
Institut za
Ispitivanje
Materijala 5.7 7 C 4.34 45.5 0.04 16.17
NE of
Banja
Luka
Bosnia
&herzegovina 1981
Banja Luka-
Borik 9 5.7 7 C 3.551 34.5 0.1 18.52
Sicilia-
Orientale Sicily, Italy 1990
Catania-
Piana 5.6 24 C 2.483 25 0.18 43.43
Basso
Tirreno Sicily, Italy 1978
Patti-
Cabina
Prima 6 18 C 1.585 10.3 0.26 30.86
Pyrgos
Southern
Greece 1993
Pyrgos-
Agriculture
Bank 5.4 10 C 4.256 22.7 0.3 25.61
Umbria
Marche Central Italy 1997 Colfiorito 5.7 3 C 3.382 11 0.34 44.46
* Epicentral distance
** Peak horizontal acceleration
***Duration
124
Chapter 3
Figure 3.24 5% damped spectra for selected real ground motion records, and their average
compared to code design spectrum
3.3.2.5.4 Post-processing of output
For each time-history run, the following output was extracted:
1. Shear force in the x-direction at the base of the six columns of the 2-D structure is
extracted from ZEUS-NL and summed up in a Microsoft excel spreadsheet at every
time step. The absolute maximum of this total base shear is calculated and tabulated.
This is the base shear for this particular run and particular earthquake. The same
procedure is applied for getting the base moment.
2. X-Displacement of the right-most column at each floor level node is extracted from
ZEUS-NL into a Microsoft excel spreadsheet. The relative displacement for each
node of this column is calculated as the difference between each particular nodes
displacement and the base nodes displacement, at each time step. The absolute
maximum is calculated for each node relative displacement and tabulated.
3. Using the same extracted results of node displacements at each time interval, the
inter-storey drift is calculated by subtracting each nodes displacement from that of
the node in the floor below it, at each time step. The absolute maximum is calculated
for each storey relative drift and tabulated.
125
Chapter 3
For each suite of 7 earthquakes (seven groups of results), the results are averaged among
seven to get:
1. One value of base shear (linear F
e
; nonlinear F
i
)
2. One value of base moment (linear M
e
; nonlinear M
i
)
3. One group of values for node displacement values at each floor n (linear Ae
n
;
nonlinear Amaxi
n
)
4. One group of values for inter-storey drift values at each floor (only nonlinear IDi
n
)
Sample of the post-processed results for case studies are presented in Appendix G.
3.4 ASSESSMENT METHODOLOGY
Because the aim of this study is to develop numbers to be used by designers, the
assessment of a rational value for DAF has to be based on the following criteria
- Generality of application.
- Ease and practicality of application in the design office
- Enveloping all possible scenarios.
In accord with current code provisions that specifies the DAF as a factor of the force reduction
factor (R in the Egyptian code), the evaluation of the DAF factor to be used for design
purposes was evaluated through the ratio of DAF to FRF, rather than DAF alone. ECP-
201(2008) provides this ratio as 0.7, as shown in equation (2.9) provided in Chapter 2.
The steps conducted in the assessment stage were as follows:
1. Displacement results at each floor n of the right-most column were extracted from
equivalent static method (de). The code estimated maximum displacement was calculated,
ds=0.7Rde, and tabulated in Appendix F.
2. Displacement output extracted from linear time-history analysis: A
e
, and tabulated in
Appendix G.
126
Chapter 3
3. Displacement output extracted from nonlinear time-history analysis: A
maxi
tabulated in
Appendix G.
4. Story drift profiles were plotted for each building using three values: A
maxi
(maximum
displacement from inelastic analysis), d
s
, (code estimate of maximum displacement =0.7 R
de) and d
sm
(code estimate of maximum displacement without lower bounds on base shear
value). The three patterns were compared.
5. Inter-storey drift profiles were also plotted for the three cases and compared.
6. Computation of displacement amplification factor (DAF) to force reduction factor (FRF):
From Figure 2.13, realizing that
=
(3.14)
Therefore this ratio was calculated using three methods as shown in Table4.2, that differ in
the value and analysis method used to compute the inelastic and elastic displacement.
These ratios are summarized in Table3.13, yielding the following equations:
Table 3.13 Definitions of the different ratios used in assessment of DAF/ FRF factor
Assessment ratio Displacement
value measured
Assessment
location
Analysis method
for obtaining
inelastic
displacement
Analysis method
for obtaining
elastic
displacement
(DAF/FRF)
computed
Storey drift All floors Nonlinear THA
(Amaxi)
Code estimate
(= Rx de)
(DAF/FRF)
actual
Storey drift Roof Nonlinear THA
(Amaxi)
THA ( Ae)
(DAF/FRF)
ID
Inter-storey drift All floors Nonlinear THA
(Amaxi)
Code estimate
(= Rx de)
Assessment ratio (1)
=
=1
(3.15)
where,
,
: is the maximum inelastic displacement as output from nonlinear time-history
analysis at floor level i
,
.(3.16)
,
: is the maximum inelastic displacement as output from nonlinear time-
history analysis at the roof level
=
=1
..(3.17)
Where,
IDI
i
: is the maximum instantaneous inter-storey drift as output from nonlinear time-
history analysis and as calculated 3.3.2.5.4 (3) of this report, at floor level i
L
i
n
e
a
r
T
H
A
B
a
s
e
c
a
s
e
s
c
e
n
a
r
i
o
s
APPENDIX G
G-4
T
a
b
l
e
G
.
2
S
t
o
r
e
y
d
r
i
f
t
r
e
s
u
l
t
s
(
m
m
)
L
i
n
e
a
r
T
H
A
B
a
s
e
c
a
s
e
s
c
e
n
a
r
i
o
s
C
o
n
t
d
APPENDIX G
G-5
T
a
b
l
e
G
.
3
S
t
o
r
e
y
d
r
i
f
t
r
e
s
u
l
t
s
(
m
m
)
-
N
o
n
l
i
n
e
a
r
T
H
A
-
B
a
s
e
c
a
s
e
s
c
e
n
a
r
i
o
s
APPENDIX G
G-6
T
a
b
l
e
G
.
3
S
t
o
r
e
y
d
r
i
f
t
r
e
s
u
l
t
s
(
m
m
)
-
N
o
n
l
i
n
e
a
r
T
H
A
-
B
a
s
e
c
a
s
e
s
c
e
n
a
r
i
o
s
C
o
n
t
d
APPENDIX G
G-7
T
a
b
l
e
G
.
4
S
t
o
r
e
y
d
r
i
f
t
r
e
s
u
l
t
s
(
m
m
)
-
N
o
n
l
i
n
e
a
r
T
H
A
-
R
e
f
i
n
e
d
l
o
a
d
i
n
g
s
c
e
n
a
r
i
o
s
APPENDIX G
G-8
T
a
b
l
e
G
.
5
S
t
o
r
e
y
d
r
i
f
t
r
e
s
u
l
t
s
(
m
m
)
-
N
o
n
l
i
n
e
a
r
T
H
A
-
R
e
-
p
r
o
p
o
r
t
i
o
n
e
d
c
a
s
e
s
(
h
i
g
h
e
r
s
t
i
f
f
n
e
s
s
)
APPENDIX G
G-9
T
a
b
l
e
G
.
6
S
t
o
r
e
y
d
r
i
f
t
r
e
s
u
l
t
s
(
m
m
)
-
N
o
n
l
i
n
e
a
r
T
H
A
-
D
i
f
f
e
r
e
n
t
d
u
r
a
t
i
o
n
s
e
a
r
t
h
q
u
a
k
e
l
o
a
d
i
n
g
APPENDIX G
G-10
T
a
b
l
e
G
.
7
S
t
o
r
e
y
d
r
i
f
t
r
e
s
u
l
t
s
(
m
m
)
-
N
o
n
l
i
n
e
a
r
T
H
A
-
D
i
f
f
e
r
e
n
t
s
t
e
e
l
m
o
d
e
l
s
APPENDIX G
G-11
T
a
b
l
e
G
.
8
S
t
o
r
e
y
d
r
i
f
t
r
e
s
u
l
t
s
(
m
m
)
-
N
o
n
l
i
n
e
a
r
T
H
A
-
M
e
a
n
v
e
r
s
u
s
c
h
a
r
a
c
t
e
r
i
s
t
i
c
m
a
t
e
r
i
a
l
p
r
o
p
e
r
t
i
e
s
APPENDIX G
G-12
T
a
b
l
e
G
.
9
S
t
o
r
e
y
d
r
i
f
t
r
e
s
u
l
t
s
(
m
m
)
-
N
o
n
l
i
n
e
a
r
T
H
A
-
S
o
f
t
-
f
i
r
s
t
s
t
o
r
e
y
c
a
s
e
APPENDIX G
G-13
T
a
b
l
e
G
.
1
0
I
n
t
e
r
-
s
t
o
r
e
y
d
r
i
f
t
r
a
t
i
o
(
%
)
-
N
o
n
l
i
n
e
a
r
T
H
A
-
B
a
s
e
c
a
s
e
s
c
e
n
a
r
i
o
s
APPENDIX G
G-14
T
a
b
l
e
G
.
1
1
I
n
t
e
r
-
s
t
o
r
e
y
d
r
i
f
t
r
a
t
i
o
(
%
)
-
N
o
n
l
i
n
e
a
r
T
H
A
-
R
e
f
i
n
e
d
l
o
a
d
i
n
g
s
c
e
n
a
r
i
o
s
APPENDIX G
G-15
T
a
b
l
e
G
.
1
2
I
n
t
e
r
-
s
t
o
r
e
y
d
r
i
f
t
r
a
t
i
o
(
%
)
-
N
o
n
l
i
n
e
a
r
T
H
A
-
R
e
-
p
r
o
p
o
r
t
i
o
n
e
d
c
a
s
e
s
(
h
i
g
h
e
r
s
t
i
f
f
n
e
s
s
)
APPENDIX G
G-16
T
a
b
l
e
G
.
1
3
I
n
t
e
r
-
s
t
o
r
e
y
d
r
i
f
t
r
a
t
i
o
(
%
)
-
N
o
n
l
i
n
e
a
r
T
H
A
-
D
i
f
f
e
r
e
n
t
d
u
r
a
t
i
o
n
s
e
a
r
t
h
q
u
a
k
e
l
o
a
d
i
n
g
APPENDIX G
G-17
T
a
b
l
e
G
.
1
4
I
n
t
e
r
-
s
t
o
r
e
y
d
r
i
f
t
r
a
t
i
o
(
%
)
-
N
o
n
l
i
n
e
a
r
T
H
A
-
D
i
f
f
e
r
e
n
t
s
t
e
e
l
m
o
d
e
l
s
APPENDIX G
G-18
T
a
b
l
e
G
.
1
5
I
n
t
e
r
-
s
t
o
r
e
y
d
r
i
f
t
r
a
t
i
o
(
%
)
-
N
o
n
l
i
n
e
a
r
T
H
A
-
M
e
a
n
v
e
r
s
u
s
c
h
a
r
a
c
t
e
r
i
s
t
i
c
m
a
t
e
r
i
a
l
p
r
o
p
e
r
t
i
e
s
APPENDIX G
G-19
T
a
b
l
e
G
.
1
6
I
n
t
e
r
-
s
t
o
r
e
y
d
r
i
f
t
r
a
t
i
o
(
%
)
-
N
o
n
l
i
n
e
a
r
T
H
A
-
S
o
f
t
-
f
i
r
s
t
s
t
o
r
e
y
c
a
s
e
Appendix H
Roof displacement time history records for the four
prototype buildings in the three zones
APPENDIX H
H-1
(a) Zone 1 (b) Zone 3 (c) Zone 5B
ART-EQ1
ART-EQ2
ART-EQ3
ART-EQ5
ART-EQ7
ART-EQ8
ART-EQ9
Figure H.1 Top displacement of the one-storey building in zones (a)1; (b)3 and (c)5B
APPENDIX H
H-2
(a) Zone 1 (b) Zone 3 (c) Zone 5B
ART-EQ1
ART-EQ2
ART-EQ3
ART-EQ5
ART-EQ7
ART-EQ8
ART-EQ9
Figure H.2 Top displacement of the four-storey building in zones (a)1; (b)3 and (c)5B
APPENDIX H
H-3
(a) Zone 1 (b) Zone 3 (c) Zone 5B
BB
ART-EQ1
ART-EQ2
ART-EQ3
ART-EQ5
ART-EQ7
ART-EQ8
ART-EQ9
Figure H.3 Top displacement of the seven-storey building in zones (a)1; (b)3 and (c)5B
APPENDIX H
H-4
(a) Zone 1 (b) Zone 3 (c) Zone 5B
ART-EQ1
ART-EQ2
ART-EQ3
ART-EQ5
ART-EQ7
ART-EQ8
ART-EQ9
Figure H.4 Top displacement of the ten-storey building in zones (a)1; (b)3 and (c)5B
Appendix J
Parametric study case: re-proportioned for higher
stiffness (Zone 5B)
APPENDIX J
J-1
F
i
g
u
r
e
J
.
1
D
e
t
a
i
l
i
n
g
i
n
e
l
e
v
a
t
i
o
n
a
n
d
c
r
o
s
s
s
e
c
t
i
o
n
s
o
f
b
e
a
m
o
f
t
h
e
o
n
e
-
s
t
o
r
e
y
f
r
a
m
e
i
n
z
o
n
e
5
B
(
h
i
g
h
e
r
s
t
i
f
f
n
e
s
s
c
a
s
e
s
t
u
d
y
)
APPENDIX J
J-2
F
i
g
u
r
e
J
.
2
D
e
t
a
i
l
i
n
g
i
n
e
l
e
v
a
t
i
o
n
a
n
d
c
r
o
s
s
s
e
c
t
i
o
n
s
o
f
b
e
a
m
o
f
t
h
e
f
o
u
r
-
s
t
o
r
e
y
f
r
a
m
e
i
n
z
o
n
e
5
B
(
h
i
g
h
e
r
s
t
i
f
f
n
e
s
s
c
a
s
e
s
t
u
d
y
)
APPENDIX J
J-3
F
i
g
u
r
e
J
.
3
D
e
t
a
i
l
i
n
g
i
n
e
l
e
v
a
t
i
o
n
a
n
d
c
r
o
s
s
s
e
c
t
i
o
n
s
o
f
b
e
a
m
s
o
f
t
h
e
s
e
v
e
n
-
s
t
o
r
e
y
f
r
a
m
e
i
n
z
o
n
e
5
B
(
h
i
g
h
e
r
s
t
i
f
f
n
e
s
s
c
a
s
e
s
t
u
d
y
)
APPENDIX J
J-4
F
i
g
u
r
e
J
.
3
D
e
t
a
i
l
i
n
g
i
n
e
l
e
v
a
t
i
o
n
a
n
d
c
r
o
s
s
s
e
c
t
i
o
n
s
o
f
b
e
a
m
s
o
f
t
h
e
s
e
v
e
n
-
s
t
o
r
e
y
f
r
a
m
e
i
n
z
o
n
e
5
B
(
h
i
g
h
e
r
s
t
i
f
f
n
e
s
s
c
a
s
e
s
t
u
d
y
)
(
c
o
n
t
'
d
)
APPENDIX J
J-5
F
i
g
u
r
e
J
.
4
D
e
t
a
i
l
i
n
g
i
n
e
l
e
v
a
t
i
o
n
a
n
d
c
r
o
s
s
s
e
c
t
i
o
n
s
o
f
b
e
a
m
s
o
f
t
h
e
t
e
n
-
s
t
o
r
e
y
f
r
a
m
e
i
n
z
o
n
e
5
B
(
h
i
g
h
e
r
s
t
i
f
f
n
e
s
s
c
a
s
e
s
t
u
d
y
)
APPENDIX J
J-6
F
i
g
u
r
e
J
.
4
D
e
t
a
i
l
i
n
g
i
n
e
l
e
v
a
t
i
o
n
a
n
d
c
r
o
s
s
s
e
c
t
i
o
n
s
o
f
b
e
a
m
s
o
f
t
h
e
s
e
v
e
n
-
s
t
o
r
e
y
f
r
a
m
e
i
n
z
o
n
e
5
B
(
h
i
g
h
e
r
s
t
i
f
f
n
e
s
s
c
a
s
e
s
t
u
d
y
)
(
c
o
n
t
'
d
)
APPENDIX J
J-7
F
i
g
u
r
e
J
.
4
D
e
t
a
i
l
i
n
g
i
n
e
l
e
v
a
t
i
o
n
a
n
d
c
r
o
s
s
s
e
c
t
i
o
n
s
o
f
b
e
a
m
s
o
f
t
h
e
s
e
v
e
n
-
s
t
o
r
e
y
f
r
a
m
e
i
n
z
o
n
e
5
B
(
h
i
g
h
e
r
s
t
i
f
f
n
e
s
s
c
a
s
e
s
t
u
d
y
)
(
c
o
n
t
'
d
)
APPENDIX J
J-8
Figure J.5 Column cross section detailing for Zone 5-B re-proportioned case buildings